← ALL RESOURCES
FREE RESOURCE · PE SPONSORS · PORTFOLIO MANAGEMENT TEAMS · 2026 EDITION

Technology exit readiness checklist.

What to fix in the technology and AI of a business 24, 18, 12, 6 and 3 months before a sale, so the buyer's due diligence finds nothing that moves the price. 97 items across 15 workstreams, each with the evidence buyers ask for, the time and cost to fix, an owner and a priority, plus a timeline, a vendor due diligence guide and a live Excel tracker.

FORPE sponsors, operating partners, portfolio CEOs, CFOs and CTOs, owner-managers 12 to 24 months from a sale
UPDATED25 October 2026
VERSION1.0
FORMATSPDF, XLSX
COSTFree. No sign-up.
RELATES TOExit readiness

KEY TAKEAWAYS

  • The Anyway Ventures technology exit readiness checklist sets out 97 items across 15 workstreams, timed back from the launch of a sale process at 24, 18, 12, 6 and 3 months.
  • Before a sale, the slowest items set the clock: reducing key-person dependency takes six to 18 months, AI results need six to twelve months of measured data, and cloud savings need twelve months to show.
  • Fix the cheap, alarming items first, such as admin access held by one person, missing IP assignments, open pen test findings and open-source licence conflicts, because buyers turn them into conditions or indemnities.
  • EY's 2026 exit readiness study found 86% of GPs said exit preparation improved valuations, with the best results where preparation started 12 to 24 months before a sale.
  • A problem the seller found, costed and disclosed is a line in the report, but the same problem found by the buyer is a negotiation, so cost and disclose fixes that cannot finish before launch.

This is a complete technology exit readiness checklist: 97 things a sponsor and portfolio management team should fix, evidence or disclose in the technology and AI of a business before a sale, organised into 15 workstreams and timed back from the launch of the process at 24, 18, 12, 6 and 3 months. For every item it sets out what the buyer's technical adviser checks, the evidence to have ready, how long the fix takes, what it costs, who owns it and how much it matters. It is written for PE sponsors, operating partners and portfolio CEOs, CFOs and CTOs who want the buyer's due diligence to find nothing that moves the price, and the Excel companion turns it into a working tracker with a live readiness score.

86%of GPs said exit preparation improved valuations; starting 12 to 24 months out worked best. EY, 2026
18%of GPs named AI as a challenge in preparing companies for sale, up from 7% a year earlier. EY, 2026
68%of 947 audited commercial codebases had open-source licence conflicts, up from 56%. Black Duck, 2026
9%of PE leaders were very confident they could pass an AI governance audit in 90 days. Grant Thornton, 2026

1. Executive summary

  1. Count back from launch, not from completion. Every date here is measured from the day the information memorandum and data room go to bidders. Completion is usually three to six months later, which is too late to fix anything. The clock is set by the slowest items: key-person dependency (six to 18 months), AI results (six to twelve months of measured data) and cloud savings (twelve months to show in the last twelve months' accounts).
  2. Do the cheap, alarming items first. Admin access held by one person, missing IP assignments, open pen test findings and open-source licence conflicts cost little to fix 18 months out and a great deal once a buyer finds them. They routinely become conditions of completion or specific indemnities.
  3. Fix, cost and disclose, or leave alone. Not everything should be fixed. A re-platform that cannot finish before launch should be costed and disclosed, not started. A problem you have found, costed and disclosed is a line in the report; the same problem found by the buyer is a negotiation.
  4. Evidence beats assertion. An item is not done until the document a buyer will ask for exists, predates the process and is filed. The tracker scores "fixed, evidence not filed" at 70% for that reason.
  5. The AI story needs a year of data. Buyers credit production usage, measured margin effect, clear data rights and governance they can check. Use cases you want credited need to be live 12 to 18 months before launch so six to twelve months of results exist. Grant Thornton found only 9% of PE leaders very confident of passing an AI governance audit in 90 days.
  6. KPIs are technology work. EY found 60% of GPs still named a solid set of data and KPIs as the hardest finance challenge in exit preparation. Revenue retention, cohort and feature usage data come out of product systems, and buyers test them against the accounts.
  7. Prepare the people as well as the platform. In EY's 2026 study, the first thing exited management teams said they would do differently was prepare the management team better. Run a mock technical due diligence six to eight weeks before launch.
  8. Use vendor due diligence deliberately. A readiness baseline 24 to 18 months out finds the problems; a vendor technical due diligence report two to four months out, with reliance for the eventual buyer, presents them. Running the first makes the second short and clean.

2. How to use this checklist

Who it is for. The sponsor's deal or operating partner owns the plan and the timetable. The portfolio CEO owns delivery, with the CTO, CFO and whoever handles legal and data protection owning individual items. Owner-managers selling without a sponsor can use it the same way, with the CEO in the sponsor's seat.

When to use it. Ideally at 24 months before the planned launch, as the frame for a readiness baseline. It still works at 12 months, but some items move from "fix" to "cost and disclose". At six months it becomes a disclosure and data room checklist.

How long it takes. A first pass, scoring every item from what management already knows, takes a CTO and CFO about a day. A buyer-grade baseline that tests each item against evidence takes two to three weeks with outside help. After that, a monthly review of 30 to 60 minutes with the sponsor keeps it moving.

Inputs you need. Repository access and commit history, the last two pen test reports, 24 months of cloud bills, contracts with employees, contractors, customers and key suppliers, the KPI pack and the definitions behind it, and any data on AI usage and cost.

How to read each item. Every item has an ID (IP-01 and so on), which matches the Excel tracker, and a final column with four things:

Field Meaning
When Start month and done-by month, counted back from launch. M18 to M6 means start 18 months before launch and finish by six months before.
Effort Typical elapsed time to fix: days, weeks, months, or a range.
Cost band £ under £5k; ££ £5k to £25k; £££ £25k to £100k; ££££ over £100k. External spend plus internal time at a loaded engineering cost of about £85k a year. Indicative for a business of 50 to 500 staff.
Owner, priority The default owner, and how a buyer's adviser would rate the item if it were left: Critical (could stop the deal or become a condition of completion), High (likely to change price or terms), Medium (100-day plan or minor warranty), Low (note for the file).

The Excel companion. The workbook holds every item on a Tracker tab with dropdowns for status, priority, cost band and owner, a due date calculated from your launch date, and columns for evidence location and notes. A Dashboard rolls the statuses into a weighted readiness score by workstream with RAG ratings, open critical items, overdue items and a launch gate. A Timeline tab draws the plan as a Gantt chart by month, coloured by status, and a Data room tab tracks the technology folders. Example tabs show the whole thing filled in for the fictional business in the worked example.

3. How a buyer reads your technology

A buyer's technical adviser is not marking you against best practice (the requests they work from are in our technical due diligence checklist). They are trying to answer three questions for an investment committee: does the business own what it sells, can the technology carry the plan the buyer is paying for, and what will it cost to put right? Every finding is costed and routed somewhere: a price adjustment or debt-like item, a condition of completion, a specific warranty or indemnity, a structural protection such as retention or deferred consideration, the 100-day plan, or the file. That structure is set out in what a technical due diligence report should contain, and it is the right lens for preparation too. The question for each item on this list is which route it would take if a buyer found it, and whether you can move it to "the file" before they do.

That gives three choices for every item.

Choice When it is right Example
Fix and evidence The fix finishes, and is evidenced, before launch IP assignments, admin access, pen test findings, cloud waste
Cost and disclose The fix cannot finish before launch, or would make the business look worse half-done A re-platform, an ISO 27001 programme started late, a framework upgrade with a year to run
Leave alone The work does not change what a buyer would pay Cosmetic refactoring, features built to dress the roadmap, certifications no customer asks for

Insurers are reading technology more closely too. Warranty and indemnity insurers expect the buyer to test vendor reports rather than accept them, as Browne Jacobson's guidance on W&I insurance and sell-side due diligence explains, and in July 2026 the International Underwriting Association's Transactional Liability Committee warned that W&I risk analysis frequently skips penetration testing and technical validation. Thin technical evidence now tends to come back as an exclusion in the W&I policy as well as a question from the buyer.

4. The timeline by months to launch

The table is the plan on one page. The item IDs refer to the checklist that follows, and the score targets match the Excel dashboard.

Window What happens Key items Gate at the end
M24 to M18 Readiness baseline. Score every item against evidence, cost each finding, agree the fix, disclose or leave decision with the board. Start the slow items immediately. IP-01 to IP-03, SE-05, KP-01, KP-02, TD-02, TD-03, AI-03 baselines, SE-04 decision Board-approved plan; admin access and founder IP fixed; re-platform decision minuted. Score target 40% by M18.
M18 to M12 Fix the cheap, high-impact items. Run the open-source scan and remove cloud waste early enough to show in the LTM. Get AI use cases live with logging. OS-01, OS-02, CC-01 to CC-03, PM-01, AI-01, AI-02, AI-04, DP-01 to DP-03, KP-05, KP-06 Scan done and copyleft plan agreed; cloud savings live; KPI definitions reconciled; AI live with usage logs. Score target 60% by M12.
M12 to M6 Prove it. Pen test with time to remediate and retest. Load test against the plan. Write the runbooks and test them. Let AI and cost results accrue. Decide on vendor due diligence. SE-01 to SE-03, AR-01 to AR-06, DO-01 to DO-04, KP-03, KP-04, CT-01, CT-02, AI-05 to AI-08, AC-01 to AC-04 Pen test findings closed or dated; retention agreed with the sponsor; VDD adviser shortlisted. Score target 80% by M6.
M6 to M3 Assemble. Appoint the VDD adviser, build the data room technology folders, write the known issues list, draft the AI page of the IM. DR-01, DR-04, DR-05, AI-09, OS-06, DP-07, KP-07 VDD fieldwork under way; data room folders populated; mock due diligence booked. Score target 90% by M3.
M3 to launch Test and rehearse. VDD fieldwork and report, mock due diligence and management rehearsals, code access arrangements, final rescans. DR-03, DR-06, DR-07, SE-02 retest, OS-01 rescan Overall 95%, no open critical items, no red workstream.
Launch to signing Answer questions, keep evidence current, run the bring-down if the process runs long. DR-04, DR-07 Q&A turnaround held; VDD bring-down letter if needed.

Why it matters. Three items set the critical path and none can be compressed by spending more: a second engineer who genuinely knows each critical system (KP-02), six to twelve months of AI results against a baseline (AI-02, AI-03), and twelve months of lower cloud cost in the trading history (CC-03). If you are reading this at M12, start those three this month and accept that some of the rest moves to "cost and disclose".

5. The checklist by workstream

Fifteen workstreams, in roughly the order a buyer's adviser reaches them. Each section opens with what usually goes wrong, then the items.

IP and ownership

The first question in every technical due diligence is whether the company owns the code it sells. Under section 11 of the Copyright, Designs and Patents Act 1988, copyright in work an employee creates in the course of employment belongs to the employer, but code written by contractors, agencies and founders before incorporation does not unless it has been assigned in writing. The gaps are almost always the agency that built version one and early freelancers, and a former contractor who learns the business is being sold has every reason to drive a hard bargain. Find them at M24.

Item What buyers check Evidence to prepare When, effort, owner
IP-01 Employee IP assignments Every current and former employee who wrote code, designs or content has terms assigning IP to the company IP register mapping every repository author to a signed contract; template clause; gaps list M24 to M12
Weeks · £ · Legal
Critical
IP-02 Contractor and agency assignments Freelancers and the agency that built version one assigned their IP in writing, not just a licence Signed assignments or confirmatory assignments; invoices matched to agreements M24 to M12
Weeks to months · ££ · Legal
Critical
IP-03 Founder and pre-incorporation code Code a founder wrote before incorporation or while a contractor has been assigned to the company Founder deed of assignment; date of incorporation against first commit date M24 to M18
Weeks · £ · Legal
Critical
IP-04 Trade marks, domains and app store accounts Product names registered in key markets; domains and app listings held by the company, not a person Trade mark schedule; registrar and app store account owner screenshots; renewal dates M18 to M6
Weeks · £ · Legal
High
IP-05 Acquired IP and chain of title Code or products bought in earlier acquisitions came across cleanly with warranties Asset purchase agreements, assignments and disclosure letters from prior deals M18 to M6
Weeks · £ · Legal
Medium
IP-06 Inbound commercial licences Licensed SDKs, fonts, data sets and embedded components permit the current use and survive a sale Register of inbound licences with term, scope, change-of-control and transfer terms M18 to M6
Weeks · £ · CTO
High

Tip. Build the IP register from the commit history, not from the HR file. Export every author email from every repository, including the archived ones, and match each to a contract. The list of unmatched names is your work list.

Open source and licences

This is more common than most CTOs think. Black Duck's 2026 Open Source Security and Risk Analysis, based on 947 commercial codebases from audits that included 197 M&A transactions, found licence conflicts in 68% of them, up from 56% the year before, and components four or more years out of date in 92%. Replacing or isolating a component is straightforward with a year in hand and fraught under a deadline, so run the scan at M18, not M3.

Item What buyers check Evidence to prepare When, effort, owner
OS-01 Software composition analysis and SBOM A scan of every repository, including mobile apps and infrastructure code, and a current bill of materials SCA report less than three months old; SBOM in SPDX or CycloneDX format; repository list M18 to M6
Weeks · ££ · CTO
Critical
OS-02 Copyleft in distributed or hosted code GPL in distributed software, AGPL or SSPL in hosted services, and whether obligations have been triggered Licence findings with remediation status; architecture note showing isolation where kept M18 to M6
Weeks to months · £££ · CTO
Critical
OS-03 Outdated and unmaintained components Components years out of date or with no maintainer, and frameworks out of security support Upgrade backlog with versions, end-of-support dates and planned completion M18 to M3
Months · £££ · CTO
High
OS-04 Open-source policy and approval A written policy, an approved licence list and a gate that stops new conflicts entering Policy; CI scan configuration; approval log; OpenChain ISO/IEC 5230 alignment if claimed M12 to M6
Weeks · £ · CTO
Medium
OS-05 Source-available and relicensed components Components moved from open-source to source-available licences that restrict competing or hosted use List of affected components, licence version in use, and the replacement or licence decision M12 to M6
Weeks · ££ · CTO
Medium
OS-06 Attribution and notice files Notices and licence texts shipped with distributed apps and on-premise software NOTICE files; in-app acknowledgements screen; release checklist M6 to M3
Days · £ · CTO
Low

Red flag. A modified AGPL component inside a hosted product, or GPL code inside software the business ships to customers, with nobody aware of it until the buyer's scan. Expect a specific indemnity and remediation before completion.

AI-generated code

This is the newest workstream and the one most sellers have not prepared for. Black Duck's same report found about 85% of organisations using AI coding assistants but only 54% evaluating AI-generated code for IP and licensing risk, and 76% of companies that prohibit AI coding tools acknowledging that developers use them anyway. Buyers' advisers now ask which tools were used, on what terms, and what stops licensed code being reproduced without its licence. The fixes are mostly settings and policy, which makes them cheap at M12.

Item What buyers check Evidence to prepare When, effort, owner
AC-01 AI coding tool inventory and terms Which assistants engineers use, on which licence tier, and whether vendors may train on company code Tool register; enterprise agreements; settings showing training opt-out; seat list M12 to M6
Weeks · £ · CTO
High
AC-02 Provenance controls for generated code Duplicate-detection or public-code filters switched on, and scans for matches with licensed code Admin settings export; SCA snippet-match results; conditions of any vendor IP indemnity M12 to M6
Weeks · £ · CTO
High
AC-03 Review standard for AI-written code Human review and tests required before merge, whoever or whatever wrote the change Branch protection rules; pull request template; coverage trend for 12 months M12 to M6
Weeks · £ · CTO
Medium
AC-04 Secrets and customer data in prompts Rules and controls stopping credentials and personal data being pasted into assistants Policy clause; secret-scanning reports; data loss prevention settings where used M12 to M6
Weeks · £ · CTO
High
AC-05 Share of code that is AI-generated An honest estimate of how much of the codebase was generated, and in which areas Method note and estimate by repository; any vendor usage analytics M6 to M3
Days · £ · CTO
Low

Security and pen testing

Buyers ask for the last two penetration test reports and what happened to the findings. Critical or high findings left open for a year tell a buyer more about the engineering culture than any certificate. On certification, choose by timetable and buyer pool. Cyber Essentials Plus, the NCSC-backed UK scheme with an independent technical audit, can be achieved in weeks and, alongside a clean pen test, answers most mid-market buyers. ISO 27001 and SOC 2 Type II both need many months of operating evidence before the audit, so start them at M24 or not at all: a programme started six months before launch reads as a scramble. If enterprise customers or a US buyer pool expect SOC 2, decide at the baseline.

Item What buyers check Evidence to prepare When, effort, owner
SE-01 Independent penetration test A test of the production application and infrastructure by a CREST or equivalent tester in the last 12 months Last two reports; scope letters; tester credentials M12 to M6
Weeks · ££ · CTO
Critical
SE-02 Remediation and retest of findings Critical and high findings closed and retested, with open items owned and dated Findings tracker; retest letter; vulnerability SLA policy and performance M12 to M3
Weeks · ££ · CTO
Critical
SE-03 Cyber Essentials Plus Current UK baseline certification, independently tested, renewed annually Certificate and scope; assessor report M12 to M6
Weeks · £ · IT lead
High
SE-04 ISO 27001 or SOC 2 Type II Whether enterprise customers or the buyer pool expect a certified ISMS or an audited control report Certificate or report, scope statement, statement of applicability or bridge letter M24 to M6
9 to 18 months · £££ · CTO
Medium
SE-05 Identity and privileged access MFA everywhere, single sign-on, leaver process, privileged access limited and reviewed Access review records; MFA coverage report; leaver checklist samples M24 to M12
Weeks · £ · IT lead
Critical
SE-06 Incident history and response plan Every security incident in three years, whether reportable ones were reported, and a tested plan Incident log; ICO reports if any; plan; tabletop exercise record M18 to M6
Weeks · £ · CTO
High
SE-07 Backup and disaster recovery tested Backups restored by someone other than their author, against stated recovery targets Restore test records with dates and timings; RTO and RPO; DR plan M12 to M6
Weeks · ££ · CTO
High
SE-08 Secure development pipeline Secret scanning, dependency scanning and protected branches in the build pipeline CI configuration; scanning dashboards; branch rules M12 to M6
Weeks · £ · CTO
Medium
SE-09 Cyber insurance Cover in place, limits, claims history and what the insurer required at renewal Policy schedule; proposal form; claims history M12 to M6
Weeks · £ · CFO
Medium

Tip. Commission the pen test at M12 rather than M6. You need time to fix the findings, retest, and have the retest letter in the data room before the VDD fieldwork starts.

Data protection

Buyers check UK GDPR compliance as a liability question and, increasingly, as a precondition for the AI story, because personal data flowing to AI vendors needs a lawful basis and a transfer mechanism. Personal data breaches that meet the threshold must be reported to the ICO within 72 hours; a buyer will compare your breach log with your incident log and with what the ICO was told.

Item What buyers check Evidence to prepare When, effort, owner
DP-01 Record of processing and data map What personal data is held, where, why, on what lawful basis and for how long Article 30 record; data flow diagram; system list tied to data categories M18 to M6
Weeks to months · ££ · DPO
High
DP-02 Customer and sub-processor terms Data processing agreements with customers and suppliers, and a published sub-processor list DPA templates and signed exceptions; sub-processor list with transfer basis M18 to M6
Weeks · ££ · Legal
High
DP-03 International transfers A lawful basis for every transfer outside the UK, including to AI and hosting vendors Transfer register; IDTA or UK Addendum; transfer risk assessments M18 to M6
Weeks · ££ · Legal
High
DP-04 DPIAs for higher-risk processing Impact assessments for high-risk processing, including AI features that use personal data DPIAs with sign-off dates; screening log M12 to M6
Weeks · £ · DPO
Medium
DP-05 Breach and rights request records Personal data breaches, whether reported within 72 hours where required, and subject access performance Breach log; ICO correspondence; DSAR log with response times M12 to M6
Days · £ · DPO
High
DP-06 Retention, deletion and test data Retention applied in practice, and no live personal data in development or test environments Retention schedule; deletion job evidence; test data approach M12 to M3
Months · ££ · CTO
Medium
DP-07 Privacy notices and cookies Notices match actual processing; cookie consent meets PECR Current notices; cookie audit; consent banner configuration M6 to M3
Days · £ · DPO
Low

Architecture and scalability

The buyer is paying for a plan, and the architecture question is whether the platform can carry it. The adviser looks for evidence: a current diagram the team drew, load tests at the volumes in the plan, the incident history and the delivery metrics. DORA's software delivery metrics are the most widely used way of putting numbers on delivery, and pulling them from the pipeline for twelve months is better evidence than any maturity score.

Item What buyers check Evidence to prepare When, effort, owner
AR-01 Current architecture diagram A diagram drawn by the team, showing services, data stores, data flows and third parties Diagram at context and container level, dated; system catalogue M12 to M6
Weeks · £ · CTO
High
AR-02 Load and performance evidence Whether the platform carries the volume in the business plan, and what breaks first Load test results against plan volumes; capacity model; performance dashboards M12 to M6
Weeks · ££ · CTO
High
AR-03 Availability and incident record Uptime, serious incidents, root causes and SLA credits paid over 24 months Status page history; incident log with post-mortems; SLA credit ledger M12 to M6
Days · £ · CTO
High
AR-04 Single points of failure and resilience Single region, single database, single person or single supplier failures that stop trading Resilience review; hosting topology; remediation or accepted-risk log M18 to M6
Months · £££ · CTO
Medium
AR-05 Growth case mapped to platform limits Whether new segments, geographies, enterprise needs or acquisitions in the plan need re-architecture Short paper linking each plan assumption to the platform change it needs, with cost M12 to M6
Weeks · £ · CTO
Medium
AR-06 Delivery metrics How fast and how safely the team ships: lead time, deployment frequency, change fail rate, recovery time 12 months of DORA metrics from the pipeline, not estimates M18 to M6
Weeks · £ · CTO
Medium

Technical debt

Every codebase has debt. What moves the price is debt nobody has costed, because the buyer will cost it for you, conservatively. A register with engineer-month estimates at a stated loaded rate turns an argument into arithmetic. The decision that matters most here is the re-platform: don't start one you can't finish before launch. A half-migrated system is two systems, both partly documented, and a buyer will price the risk of the second half going wrong.

Item What buyers check Evidence to prepare When, effort, owner
TD-01 Costed technical debt register The material debt items, each with an engineer-month estimate and a cost at a stated loaded rate Register with basis of estimate; prioritisation; budget in the plan M18 to M6
Weeks · £ · CTO
High
TD-02 End-of-life frameworks and runtimes Languages, frameworks, databases or operating systems out of vendor security support Version inventory with end-of-support dates; upgrade plan or completion evidence M24 to M6
Months · £££ · CTO
Critical
TD-03 Re-platform decision Whether a rewrite or migration is under way, how far through, and whether it will finish before launch Board paper choosing finish, pause or defer, with cost, timeline and risk M24 to M18
Days · £ · CEO
High
TD-04 Test automation and release process Automated tests and deployments, and whether releases depend on manual steps by one person Pipeline configuration; coverage trend; release log M18 to M6
Months · ££ · CTO
Medium
TD-05 Legacy products and customer-specific versions Old versions or bespoke forks kept alive for a few customers, and what they cost to support Product version census; revenue and cost per legacy line; sunset plan M18 to M6
Months · ££ · CPO
Medium
TD-06 Engineering time allocation The split between new features, maintenance and support, and how it has moved 12 months of time or ticket data by category M12 to M6
Days · £ · CTO
Low

Key-person risk and team

The hardest workstream, because it can't be fixed with a document. The tests a buyer runs are set out in key person risk in due diligence: commit concentration by system, who holds admin access, documentation someone else has used, and who resolves incidents. The access item takes days; the knowledge item takes a year or more. Start the second at M24 even though it finishes last.

Item What buyers check Evidence to prepare When, effort, owner
KP-01 Admin access and credentials Critical accounts owned by the company with at least two named administrators and MFA Access register for cloud, code hosting, domains, DNS, app stores, payments and identity M24 to M18
Days · £ · CTO
Critical
KP-02 Knowledge concentration by system Commit and review history by area: billing, permissions, data model, integrations, infrastructure Concentration analysis by area over 24 months; named second engineer per critical system M24 to M6
6 to 18 months · £££ · CTO
Critical
KP-03 On-call rota and incident handling At least three people on call, and serious incidents resolved by different engineers Rota for 12 months; incident log showing resolver M12 to M6
Months · ££ · CTO
High
KP-04 Retention of key people Who is critical, what keeps them through the sale, and notice periods and covenants Key-person list; agreed retention, rollover or MIP terms; contract summaries M12 to M6
Months · ££££ · Sponsor
High
KP-05 Contractor and agency dependency Core code or operations held by contractors or an agency with no notice period or retention Contractor list with roles; conversion or knowledge-transfer plan M18 to M6
Months · ££ · CTO
High
KP-06 Technology leadership for the next owner Whether the CTO can carry the next plan and present to buyers, or a hire is needed Leadership assessment; succession or hire plan agreed with the sponsor M18 to M12
Months · £££ · Sponsor
High
KP-07 Organisation, tenure and attrition Org chart with tenure, contractors, open roles and two years of leavers Org chart; leaver list with reasons; engineering hiring record M6 to M3
Days · £ · People
Medium

Red flag. The founder CTO is about to receive a large cheque at completion, has no rollover, is the only person who has touched the billing service in two years, and is the sole owner of the cloud root account. Each of those is fixable; all four together will change the deal structure.

Documentation

Documentation speeds up diligence and shrinks the key-person finding, but only if someone other than its author has used it. Write it for the next engineer you hire. A buyer will find that version more useful than anything written for the data room.

Item What buyers check Evidence to prepare When, effort, owner
DO-01 Runbooks tested by a non-author Deploy, rollback, restore and incident runbooks that someone other than the author has followed Runbooks with last-used date and user M12 to M6
Weeks · £ · CTO
High
DO-02 Onboarding documentation What the most recent engineering hire used in their first fortnight, and what was missing Onboarding guide; new joiner feedback M12 to M6
Weeks · £ · CTO
Medium
DO-03 System catalogue and decision records Each service with its owner, language, hosting and dependencies, plus the reasons for major choices Catalogue; architecture decision records M12 to M6
Weeks · £ · CTO
Medium
DO-04 Security and IT policy set Current, approved policies that match practice, with review dates Policy index with owner, approval and review dates M12 to M6
Weeks · £ · IT lead
Medium
DO-05 API and integration documentation Public and partner API docs, and how integrations with customers' systems work API reference; integration list with volumes M6 to M3
Weeks · £ · CTO
Low

Cloud cost and unit economics

Buyers compare hosting cost growth with revenue growth and look for the gross margin effect. Flexera's 2026 State of the Cloud report, a survey of 753 cloud decision-makers, found respondents estimated 29% of their cloud spend was wasted, the first rise in five years, with AI workloads blamed. A saving removed at M18 is in the last twelve months' EBITDA by launch, where it is multiplied; the same saving removed at M3 is a run-rate adjustment the buyer's financial adviser will challenge. Capitalised development sits here too, because a buyer's quality of earnings work tests capitalised engineering time against engineering records.

Item What buyers check Evidence to prepare When, effort, owner
CC-01 Hosting cost history reconciled to the P&L 24 months of cloud and tooling spend, reconciled to the accounts, against revenue growth Monthly bills by provider; reconciliation; hosting as a share of revenue M18 to M6
Weeks · £ · CFO
High
CC-02 Cost allocation and unit cost Spend tagged by product, environment and, where possible, customer, giving cost per unit of delivery Tagging policy; unit cost report; gross margin by product M18 to M6
Months · ££ · CTO
High
CC-03 Cloud waste removed early Unused environments, oversized instances and idle storage removed in time to show in the LTM Savings log with dates and monthly run-rate effect M18 to M12
Months · ££ · CTO
High
CC-04 Commitments and minimum spends Reserved capacity, savings plans, marketplace commitments and their terms on a sale Commitment schedule with expiry and change-of-control terms M12 to M6
Days · £ · CFO
Medium
CC-05 AI inference cost in gross margin Model and AI vendor costs per use, included in cost of sales, with the trend Cost per AI transaction; vendor invoices; margin by AI feature M12 to M6
Weeks · £ · CFO
High
CC-06 Capitalised development and R&D claims Capitalised engineering time and R&D tax claims supported by engineering records Capitalisation policy; time records; project list matching claims M18 to M6
Weeks · ££ · CFO
High

Product metrics and KPIs

KPI quality is where exit readiness most often fails, and it is technology work because the numbers come out of product, billing and CRM systems. EY's 2026 study found 60% of GPs named developing a solid set of data and KPIs as their most challenging finance issue in exit preparation, down from 72% the year before but still top of the list, and noted the gap between internal tracking and the detail buyers need to underwrite the equity story. A net revenue retention figure that differs between the sales deck and the finance pack costs credibility on every other number.

Item What buyers check Evidence to prepare When, effort, owner
PM-01 KPI definitions reconciled to finance ARR, net and gross revenue retention, churn and customer counts defined once and tied to the accounts KPI dictionary; reconciliation of KPI pack to management accounts M18 to M12
Weeks · £ · CFO
Critical
PM-02 Cohort and segment retention Monthly retention by cohort, segment and product for at least three years Cohort tables from source systems; customer-level data cube M18 to M6
Weeks · ££ · CFO
High
PM-03 Feature usage analytics Which features customers use, how often, and which ones correlate with retention Product analytics exports by feature and segment; definitions M18 to M6
Months · ££ · CPO
High
PM-04 Roadmap delivery record Last 12 to 24 months of roadmap commitments against what shipped Dated roadmaps; release notes; variance commentary M12 to M6
Days · £ · CPO
Medium
PM-05 Reproducible KPI pack Board metrics generated from systems, not a spreadsheet one person maintains Data pipeline description; owner and backup owner; query definitions M12 to M6
Months · ££ · Head of data
High
PM-06 Customer health and concentration Revenue concentration, support volumes, satisfaction and at-risk accounts Top customer list; support ticket trends; health scores M12 to M6
Weeks · £ · CPO
Medium

AI equity story evidence

Buyers pay for AI that has already changed a number in the accounts, and they test it with questions like those in our AI due diligence questions. EY's study puts it plainly: "A list of pilots or isolated productivity tools may not be enough to support valuation." In Grant Thornton's July 2026 article, Buyers are asking questions about AI, Chen Liu of Grant Thornton Stax says "buyers want to see evidence that AI has been used to improve the bottom line", and notes buyers examine whether proprietary data is used to train external models and whether it is segregated from other customers' data. The four kinds of evidence that move a buyer (production usage, measured margin effect, data rights and governance) are explained in will buyers pay for your AI?. The items below turn them into a work plan, with the disruption question alongside, because a buyer will ask it whether or not you do. The AI disruption risk scorecard gives a method for AI-06.

Item What buyers check Evidence to prepare When, effort, owner
AI-01 AI inventory Every AI system and feature, the model or vendor behind it, the data it uses and its owner Inventory with owner, vendor, data categories and status M18 to M12
Weeks · £ · Head of data
High
AI-02 Production usage data Usage measured from logs over at least six months: users, volume, share of customers Raw usage exports and charts per system, with definitions M18 to M6
6 to 12 months · ££ · Head of data
Critical
AI-03 Baselines and measured margin effect A before-and-after measure on the same definition, and whether cost left the business Baseline method; monthly measurement; P&L and headcount trace M24 to M6
6 to 12 months · ££ · CFO
Critical
AI-04 Data rights Contracts allow the data use the AI story relies on; vendors may not train on company data; segregation Contract clause review; vendor terms; customer consent or notice where needed M18 to M6
Months · ££ · Legal
Critical
AI-05 AI governance Policy, tool register, training records, output error log and EU AI Act position AI policy; register; training completion; incident log; regulatory position paper M12 to M6
Weeks · £ · CEO
High
AI-06 Evaluation and quality evidence How AI output quality is tested, monitored and improved, and the error rates Evaluation sets; quality dashboards; release gates for model or prompt changes M12 to M6
Months · ££ · Head of data
Medium
AI-07 AI disruption exposure Where AI could compress pricing, demand or barriers to entry, and what management is doing Written exposure assessment with revenue at risk and response plan M18 to M6
Weeks · ££ · CEO
High
AI-08 AI key-person cover Two people who can explain and demo every AI system; prompts and integrations documented Named owners and deputies; documentation; demo script on live data M12 to M6
Months · ££ · CTO
High
AI-09 AI section of the information memorandum One page in which every AI claim points to a document in the data room Draft page with evidence references; independent challenge notes M6 to M3
Weeks · £ · CEO
Medium

Tip. If the business sells into the EU, AI-05 should state its position under the EU AI Act. Prohibited practices have applied since February 2025 and Article 50 transparency duties since August 2026; the Digital Omnibus deferred high-risk obligations for standalone systems to 2 December 2027 and for AI in regulated products to 2 August 2028, as Gibson Dunn sets out. Those dates fall inside a buyer's hold period.

Contracts and third parties

Change-of-control clauses are the classic late surprise: a contract worth 8% of revenue that the customer can terminate on a sale. Critical suppliers matter as much, now including AI model vendors whose terms, pricing and data use can change faster than the hosting contracts most registers were built around.

Item What buyers check Evidence to prepare When, effort, owner
CT-01 Change of control and assignment clauses Customer, supplier and partner contracts that terminate, need consent or reprice on a sale Contract register with change-of-control flags and the revenue affected M12 to M6
Weeks · ££ · Legal
High
CT-02 Critical supplier register Suppliers the business cannot trade without, including hosting, payments and AI model vendors Register with spend, term, exit options and concentration M12 to M6
Weeks · £ · CTO
High
CT-03 Customer technology commitments SLAs, security obligations, liability caps and audit rights promised to customers Matrix of non-standard terms by customer; SLA performance M12 to M6
Weeks · ££ · Legal
Medium
CT-04 Software licence compliance Counts of commercial licences against use, and any vendor audit history Licence entitlement and deployment reconciliation; audit correspondence M12 to M6
Weeks · ££ · IT lead
Medium
CT-05 Outsourced development and managed services Contracts with development partners and MSPs: IP, exit assistance, notice and staff transfer Contracts; exit plans; key individuals at the partner M12 to M6
Weeks · £ · Legal
Medium
CT-06 Source code escrow commitments Escrow given to customers and whether a sale or insolvency triggers release Escrow agreements, beneficiaries and deposit verification records M12 to M6
Days · £ · Legal
Low

IT and internal systems

Internal IT is rarely where value is won, but it is where cheap findings pile up: unmanaged laptops, leavers with live accounts, no DMARC. Cyber Essentials Plus covers much of it in one exercise. If the business shares systems or people with a group or sponsor, the separation cost belongs in the plan at M18, because a buyer will need a transitional services agreement and will price it.

Item What buyers check Evidence to prepare When, effort, owner
IT-01 Endpoint security and device management Managed, encrypted devices with endpoint detection and patching across the estate Device inventory; MDM and EDR coverage reports; patch compliance M12 to M6
Weeks · ££ · IT lead
High
IT-02 SaaS inventory and spend Every SaaS tool, who owns it, what data it holds and what it costs SaaS register from finance and identity data; renewal calendar M12 to M6
Weeks · £ · IT lead
Medium
IT-03 Finance, CRM and ERP systems Whether core business systems can produce the data a buyer will ask for without manual rework System map; data extracts tested against the expected request list M18 to M6
Months · ££ · CFO
Medium
IT-04 Email and domain security SPF, DKIM and DMARC in force; no lookalike or lapsed domains DNS records; DMARC reports; domain register M6 to M3
Days · £ · IT lead
Low
IT-05 Shared services and separation Systems, licences or people shared with a group or sponsor that would need a transitional services agreement Shared services list; separation plan and cost M18 to M6
Months · £££ · CFO
High
IT-06 IT support model Who supports staff systems, the MSP contract, service levels and ticket history MSP contract; service reports; internal IT org chart M6 to M3
Days · £ · IT lead
Low

Data room preparation

The data room is where preparation becomes visible. A buyer's adviser notices dates: a policy approved last month, an architecture diagram created the week the data room opened. Evidence that predates the process is believed. The known issues list (DR-05) is the most important document in the technology section, because it lets the seller decide how each problem is presented.

Item What buyers check Evidence to prepare When, effort, owner
DR-01 Technology folder index and owners A complete technology section, numbered, with a named owner per folder Folder index; owner list; upload tracker M6 to M3
Weeks · £ · CTO
High
DR-02 Evidence dated before the process Documents that existed before the sale, rather than written for it Version history; approval dates; board minutes referencing the documents M12 to M3
Weeks · £ · CEO
Medium
DR-03 Code access arrangements How the buyer's adviser will scan and review code: read-only access, scanning, clean room terms Access protocol; NDA terms; pre-run scan outputs M6 to M3
Weeks · £ · CTO
High
DR-04 Q&A process and prepared answers Who answers technology questions, how fast, and draft answers to the predictable ones Q&A owners; turnaround target; answer bank for the top 50 questions M6 to M3
Weeks · £ · CTO
High
DR-05 Known issues and disclosure list Every known technology issue, costed, with a remediation plan, ready for the disclosure letter Issue list with cost range, plan and status; agreed with legal advisers M6 to M3
Weeks · £ · CEO
Critical
DR-06 Clean team and redaction Competitively sensitive data protected when a trade buyer is bidding Clean team protocol; redaction log M3 to launch
Days · £ · Legal
Medium
DR-07 Bring-down plan How the evidence and the VDD report will be refreshed if the process runs long Refresh calendar; named owners M3 to launch
Days · £ · CEO
Low

6. Scoring readiness

The Excel dashboard scores readiness the same way for every workstream, so progress can be compared month to month and across a portfolio.

Status values. Each item takes one status. The value reflects how much of the buyer's concern it removes.

Status Value Meaning
Not started 0% No work done
In progress 40% Work under way, not finished
Fixed, evidence not filed 70% The fix is done but the document a buyer will ask for is not in the data room
Costed and disclosed 80% Deliberately not fixed before launch; costed, with a plan, and on the known issues list
Done and evidenced 100% Fixed, and the evidence is filed and dated before the process
Not applicable Excluded Does not apply to this business; drops out of the score

Weights. Critical 5, High 3, Medium 2, Low 1. A workstream's readiness is the sum of weight times status value, divided by the sum of applicable weights. The overall score uses the same formula across all items.

RAG. Green at 80% and above, amber from 60% to 79%, red below 60%. Data room preparation will show red until about M6, because its items start late; read it against the timeline, not in isolation.

Targets. 40% by M18, 60% by M12, 80% by M6, 90% by M3 and 95% at launch. The dashboard compares the current score with the target for the months remaining.

The launch gate. Launch when three things are true: overall readiness of at least 80%, no critical item still not started or in progress, and no workstream red. If the gate fails at M3, the choice is between moving items to "costed and disclosed" (and briefing the VDD adviser) or moving the launch date. Moving the date is usually cheaper than a late discovery in exclusivity.

Why it matters. "Costed and disclosed" scores 80%, not 100%, on purpose. Disclosure removes the surprise, which is most of the price risk, but the buyer will still deduct the cost. A plan full of disclosures can pass the gate; it will not get the same price as one full of fixes.

7. Preparing the management team for technical due diligence

In EY's 2026 study, the number one thing exited management teams said they would do differently in the sale process was better preparation of the management team. Technology is where that bites hardest, because the CTO is often the person in the room with the least practice at selling, and the buyer's technical adviser is usually the most sceptical person on the other side.

The sessions to expect

A buy-side technical due diligence on a mid-market software business usually runs over two to four weeks, and the management sessions follow a familiar pattern.

Session Who attends What the adviser is testing
Technology section of the management presentation, 20 to 30 minutes CEO, CTO Whether the technology story matches the equity story, and whether the CTO can carry it
Technology deep dive, two to four hours CTO, lead engineers, head of product Architecture, delivery, debt, roadmap realism, cost to run
Security and data protection session CTO or security lead, DPO Pen test history, incidents, controls, UK GDPR
Product, data and AI session with live demo CPO, head of data Whether the AI exists, what it costs per use, whether customers use it
Engineer interviews, sometimes without the CTO Two to four engineers Whether the team tells the same story as its leader
Follow-up questions and code review CTO, named deputies The detail behind earlier answers

Preparation steps

  • M6: write the fact base. A two-page technology summary and a sheet of the 30 or so numbers the team must know without looking: headcount by role, deployment frequency, change fail rate, uptime, hosting cost as a share of revenue, cost per AI transaction, test coverage, open pen test findings, debt register total, retention metrics.
  • M6: agree who answers what. One owner per topic, with a deputy. The CEO does not answer architecture questions; the CTO does not answer revenue questions.
  • M4: reconcile every claim. Read the IM, the VDD report and the management presentation side by side. Any number that appears twice must be the same number, from the same source, on the same definition.
  • M3: run a mock due diligence. Six to eight weeks before launch, have someone independent run the deep dive as a buyer's adviser would, including engineer interviews without the CTO. Record the answers that went badly and fix the evidence, not just the wording.
  • M3: rehearse the demo. On live data, by two different people, with a fallback if the network fails. A staged demo found out is worse than no demo.
  • M2: prepare engineers. Not with scripts, which advisers spot immediately. Tell them what the process is, that honesty is expected, and that "I don't know, but X does" is a good answer.
  • Launch: set up the question log. Anything not answered in a session goes into the Q&A log with an owner and a turnaround target, typically two working days.

Questions to rehearse

Question a buyer's adviser will ask Who answers What a good answer contains
"If you left tomorrow, what would break first?" CTO A specific system, the named deputy, and when the deputy last ran it alone
"Show me last year's roadmap against what shipped." CPO The document, the variance and the reason, without excuses
"What would you rebuild if you started again?" CTO The debt register, with costs, and why it hasn't been done
"What does the AI feature cost per use, and who uses it?" Head of data The unit cost, the usage share and trend, from logs
"Can customer data be used to train models?" CTO, Legal What the contracts say, what the vendors' terms say, and the segregation in place
"Tell me about your worst incident in two years." CTO What happened, what the customer impact was, the post-mortem and what changed
"What is in the code you didn't write?" CTO The SCA result, the copyleft position and the remediation status
"How much of the code was written by AI tools?" CTO An honest estimate, the tools and terms, and the review standard
"Why has hosting cost grown faster than revenue?" CTO, CFO The cause, what has been removed and the unit cost trend
"Which of these KPIs would a new owner calculate differently?" CFO The KPI dictionary and reconciliation, and any known definitional choices

Tip. Three rules for every session. Answer the question asked, then stop. Never use a number that isn't in the data room. Raise known issues before you are asked: a CTO who opens with "here are the three things you'll find, and what we've done about each" changes the tone of the whole diligence.

8. Vendor technical due diligence: when, scope and reliance

Vendor technical due diligence (VDD) is an independent review of the business's technology, commissioned by the seller and shared with bidders. It is distinct from the readiness baseline: the baseline is private, blunt and early; the VDD report is written for bidders, factual and late. EY's 2026 study excluded vendor due diligence from its definition of exit readiness, and that is the right way to think about it: the VDD report presents the preparation, it doesn't replace it. When it is worth commissioning, and what to fix before it, is covered in vendor technical due diligence.

When

Point What happens
M24 to M18 Readiness baseline, private to the seller. Two to three weeks.
M9 to M6 Decide whether to commission VDD, based on the likely buyer pool, how central technology is to value and how clean the baseline now looks.
M6 to M4 Appoint the adviser. Agree scope, the reliance terms and the timetable. Choose an adviser the likely bidders recognise.
M4 to M2 Fieldwork, typically two to four weeks for a mid-market business, then a draft, management's factual accuracy comments and a final report.
Launch Report in the data room, released to bidders under release letters. Optional Q&A session with bidders and the adviser.
Signing Reliance letter to the winning bidder and, usually, its lenders. A bring-down letter if the process has run long.

Scope

Match the scope to the questions the likely buyers will ask, not to a standard template. For a software or tech-enabled business it should cover:

Area What the VDD should conclude
Architecture and scalability Whether the platform carries the plan, what breaks first and what fixing it costs
Code, delivery and technical debt How fast and how safely the team ships, and the costed debt
Team and key-person risk Whether the product survives the departure of any one person
Security and data protection Pen test status, incident history, UK GDPR position
IP, open source and AI-generated code Whether the company owns what it sells, and the licence position from a scan
Cost to run Hosting as a share of revenue, the trend and unit cost
AI Whether the AI described in the IM exists, what it costs per use, and the disruption exposure
Known issues Each with the cost range and management's remediation plan

Keep management's commentary separate from the adviser's findings, and do not edit findings. A report buyers distrust adds nothing, and one that reads like marketing gets ignored.

Reliance letters

Bidders usually see the report before any of them has a contractual relationship with its author. They receive it under a release letter (sometimes called a non-reliance or hold-harmless letter), in which they agree they cannot claim against the adviser. The eventual buyer, and often its lenders, then receive a reliance letter, which gives them the right to rely on the report as if they had commissioned it, subject to agreed terms. Points to settle when you appoint the adviser:

  • Who may rely: the buyer, its lenders, and whether affiliates and the W&I insurer are included.
  • The liability cap, and whether it is shared across all relying parties.
  • The date the report speaks to, and whether a bring-down letter is available and at what cost.
  • Who pays for reliance: usually the seller as part of the engagement, sometimes recharged.
  • The scope and limitations section, which lenders and insurers read most carefully.

Reliance does not stop the buyer asking questions. Most buyers commission targeted confirmatory work on the areas their plan depends on, often security, scalability or AI, and W&I insurers expect them to. The realistic goal is that buyers trust the report enough to price from it and limit their own work to confirmation.

Cost

Our exit readiness readiness baseline is £15k to £25k over two to three weeks. A vendor technical due diligence report written for bidders, with reliance for the eventual buyer, is £25k to £40k over three to four weeks, depending on the size and complexity of the technology estate.

9. The data-room technology folder structure

Most data room indexes give technology one folder. It needs sixteen. Number them to fit your adviser's index; the structure below assumes technology is section 4, and the folder numbers match the evidence locations in the Excel example.

Folder Name Contents Owner
4.0 Technology overview One-page technology summary; system catalogue; org chart; glossary of internal system names CTO
4.1 IP and ownership IP register; assignment templates and signed exceptions; trade marks and domains; inbound licences Legal
4.2 Open source SCA report; SBOM; remediation tracker; open-source policy CTO
4.3 AI-generated code Coding tool register and terms; provenance settings; review standard CTO
4.4 Security Pen test reports and retests; certificates; policies; incident log; DR test records; cyber insurance CTO
4.5 Data protection Article 30 record; data map; DPAs; sub-processors; transfer register; DPIAs; breach and DSAR logs DPO
4.6 Architecture and operations Diagrams; load tests; availability and incident history; DORA metrics; resilience review CTO
4.7 Technical debt and roadmap Costed debt register; end-of-life inventory; re-platform paper; roadmap and delivery record CTO
4.8 Team Engineering org chart; tenure; contractors; on-call rota; key-person analysis People
4.9 Documentation Index of runbooks and onboarding material with last-used dates CTO
4.10 Cloud cost and unit economics 24 months of bills; reconciliation; unit cost; commitments; capitalisation policy CFO
4.11 Product metrics KPI dictionary; cohort tables; feature usage; customer health CPO
4.12 AI evidence pack AI inventory; usage exports; baselines and measurement; data rights; governance; exposure paper Head of data
4.13 Contracts and suppliers Contract register with change-of-control flags; critical suppliers; licence compliance; escrow Legal
4.14 IT and internal systems Device and SaaS inventories; Cyber Essentials scope; shared services; MSP contract IT lead
4.15 Vendor reports and Q&A VDD report; reliance and release letter templates; known issues list; Q&A log CEO

Tip. Put a one-page index at the top of each folder listing what is in it, its date and which checklist items it evidences. A buyer's adviser with an index asks fewer, better questions.

10. Worked example

Illustrative example. Calderbrook Systems Ltd is a fictional company. The business, its sponsor, the findings and the figures are invented to show how the checklist and tracker work, and are not drawn from any real business or deal. Costs assume a loaded engineering cost of £85,000 a year.

Calderbrook Systems Ltd, a fictional target, sells workforce scheduling software to UK care-home groups. It has £17.5m of revenue, of which £15.2m is recurring, EBITDA of £4.1m and 160 staff, 31 of them in product and engineering. Hambleton Equity Partners, a fictional sponsor, invested in 2021 and plans to launch a sale in October 2027. The readiness baseline ran in October 2025, at M24.

What the baseline found

The baseline scored the business at 38%. The ten findings that mattered most, and what the board decided:

Finding at M24 Decision Cost Status at M12
AWS root account registered to the founder CTO's personal email; sole administrator Fix in days Negligible Done (KP-01)
The agency that built version one in 2017 never assigned the IP Confirmatory assignment £6k paid to the agency plus £3k legal fees Signed, awaiting filing (IP-02)
Modified AGPL charting component inside the hosted reporting module Replace Two engineers for 2.5 months: 2 × £85k × 2.5 ÷ 12 = about £35k 60% done (OS-02)
Front end built on AngularJS, out of support since the end of 2021 Pause the rewrite, replace the riskiest screens, cost and disclose the rest Five engineers for 12 to 16 months: £425k to £565k Costed and disclosed (TD-02)
CTO behind 64% of changes to the scheduling engine in 24 months Second engineer paired on every release; VP Engineering hire About £150k including recruitment and overlap CTO share down to 38%; hire starts January 2027 (KP-02, KP-06)
Last pen test 14 months old, four high findings open New test and remediation £14k test and retest plus engineering time Three of four closed; retest booked (SE-01, SE-02)
Cloud cost up 41% year on year against revenue up 19% Rightsizing, unused environments, savings plans Engineering time £214k a year removed by April 2026 (CC-03)
Net revenue retention reported as 111% by sales and 104% by finance One definition, reconciled to the accounts Internal time 106% on the agreed definition (PM-01)
AI rota auto-fill feature live but unmeasured; customer contracts silent on use of data to improve the product Usage logging, baseline, new contract terms at renewal Internal time plus legal review 22% of customer sites using it and rising; new terms on renewals since January 2026 (AI-02, AI-04)
No ISO 27001; Cyber Essentials had lapsed Recertify Cyber Essentials Plus; do not start ISO 27001 before the sale About £3k Recertified July 2026; ISO 27001 roadmap costed and disclosed (SE-03, SE-04)

Two numbers explain why the sponsor started at M24. The £214k cloud saving was in place by April 2026, so by launch the last twelve months' accounts carry it in full; at an illustrative ten times EBITDA it is worth about £2.1m of enterprise value, and nobody has to argue about a run-rate adjustment. The scheduling engine dependency needed a year of paired releases before the commit history showed a real change, which no amount of money could have compressed into a pre-launch quarter.

The score at M12

By October 2026, twelve months before launch, the Example tabs in the workbook show:

Workstream Readiness RAG Open critical
IP and ownership 76% Amber 0
Open source and licences 68% Amber 1
AI-generated code 69% Amber 0
Security and pen testing 83% Green 1
Data protection 55% Red 0
Architecture and scalability 55% Red 0
Technical debt 79% Amber 0
Key-person risk and team 64% Amber 1
Documentation 69% Amber 0
Cloud cost and unit economics 74% Amber 0
Product metrics and KPIs 69% Amber 0
AI equity story evidence 49% Red 3
Contracts and third parties 89% Green 0
IT and internal systems 73% Amber 0
Data room preparation 21% Red 1
Overall 65% Amber 7

The overall score is ahead of the 60% target for M12, but the dashboard points to three things the headline hides. First, AI equity story evidence is red with three open critical items: the auto-fill usage data needs six more months to show a trend, the support triage baseline was only taken in January, and data rights depend on renewals. That is the workstream to resource now, because it cannot be rushed later. Second, data protection is red because the rota auto-fill feature sends staff names and shift patterns to a third-party model without a DPIA (DP-04) and with the transfer basis still being documented (DP-03), which also undermines the AI story. Third, the architecture work has not yet produced a load test at three times peak volume, which the buyer will ask for because the plan assumes the business doubles its customer sites. Data room preparation is red as expected at M12.

The board's actions from the M12 review: move the head of data's deputy onto AI evidence full time, complete the DPIA and transfer documentation by January, bring the load test forward to December, and shortlist VDD advisers in March for an April appointment.

11. Common mistakes

  1. Starting at six months. Six months is enough to write a story and not enough to generate the evidence for one. EY found firms that began preparation less than six months before exit reported materially weaker outcomes than those that started 12 to 24 months out.
  2. Counting from completion instead of launch. Bidders form their view in the first round from the data room. Anything fixed after launch is fixed in front of them.
  3. Starting a re-platform before a sale. A rewrite 60% done with the lead engineer on notice is priced as the worst case. Cost it and disclose it instead.
  4. Chasing a certification late. ISO 27001 or SOC 2 started at M6 won't finish and signals a scramble. Cyber Essentials Plus and a clean, retested pen test answer most mid-market buyers.
  5. Fixing without evidencing. The IP assignments were signed but are in someone's inbox; the restore test happened but nobody wrote down how long it took. Undocumented work earns little credit.
  6. Writing documents for the data room. Policies approved the month before launch and diagrams drawn for bidders are recognised as such. Evidence that predates the process is believed.
  7. Counting AI licences as adoption. A Copilot rollout to 300 staff is a cost line unless usage and output data show it changed something.
  8. Leaving AI-generated code unexamined. Buyers now ask which assistants wrote the code and on what terms. "We don't use them" is rarely true and easy to disprove.
  9. Two versions of the same KPI. A retention figure that differs between the IM and the finance pack undermines every other number. Reconcile definitions at M18.
  10. Late cloud savings presented as run rate. A saving made at M3 will be challenged as an adjustment. The same saving at M18 is simply in the accounts.
  11. Treating the CTO as the only key person. The dependency that causes trouble is often one level down: the contractor who wrote the integrations, the analyst who builds the board pack, the engineer who wrote every prompt.
  12. A VDD report that ignores the readiness findings. Commissioning a vendor report without first fixing what the baseline found produces a well-evidenced list of problems for bidders to price.
  13. Hiding things. The buyer's scan, interviews and contract review will find them. A problem found by the buyer becomes a price chip, a specific indemnity or a W&I exclusion, and every other answer is then doubted.
  14. Under-preparing management. Engineers contradicting the CTO, a demo that fails, numbers that don't match the data room. Each costs more credibility than the issue behind it.

12. Glossary

Term Meaning
Launch The day the information memorandum and data room go to bidders; the reference point for every month in this checklist
M18, M12 and so on Months before launch
VDD Vendor due diligence: an independent report commissioned by the seller for bidders
Release letter The letter under which bidders see a VDD report, agreeing they cannot claim against its author
Reliance letter The letter giving the buyer, and usually its lenders, the right to rely on a VDD report, subject to a liability cap
Bring-down letter An update confirming a report still reflects the business at a later date
SCA Software composition analysis: a scan identifying open-source components, licences and known vulnerabilities
SBOM Software bill of materials: a list of every component in the software
Copyleft Licences such as GPL and AGPL that require derived code to be released under the same terms in some circumstances
LTM Last twelve months of trading, the basis of most valuations
QoE Quality of earnings: the financial due diligence that tests EBITDA and its adjustments
NRR, GRR Net and gross revenue retention
DORA metrics Software delivery measures: change lead time, deployment frequency, change fail rate and recovery time
Bus factor The number of people who would have to leave before a system stalls
DPIA Data protection impact assessment under UK GDPR
W&I Warranty and indemnity insurance, which covers breaches of the seller's warranties
Condition of completion Something that must be done before the deal completes and money moves

13. Sources and further reading

Statistics cited in this checklist:

Other free resources worth reading, credited fairly:

Related Anyway Ventures articles: vendor technical due diligence, will buyers pay for your AI?, what a technical due diligence report should contain, key person risk in due diligence, technology due diligence checklist, technical due diligence questions, AI washing in due diligence and measuring AI ROI.

14. About this resource

Written by Anyway Ventures, the York technology and AI practice led by Brannan Coady, who has been through technical due diligence as a founder and CEO selling businesses and as an investor commissioning it. Version 1.0, October 2026. Free to use, adapt and share with attribution. Not legal, tax or investment advice.

Questions

How early should a business start preparing its technology for a sale?

Two years before the planned launch of the sale process is ideal, and 18 months is the latest point at which every item on this checklist can still be fixed rather than disclosed. The slow items set the clock: reducing key-person dependency takes six to 18 months, AI results need six to twelve months of measured data, and cloud savings need a full year to show in the last twelve months' accounts. EY's 2026 exit readiness study found the strongest outcomes where preparation began 12 to 24 months before a sale.

What do buyers look at in technical due diligence?

Whether the business owns what it sells, whether the platform can carry the plan, who holds the knowledge, how secure it is, what it costs to run, how much debt sits in the code, and whether the AI story is real. In practice they read IP assignments, open-source scan results, pen test reports, commit history, cloud bills, KPI definitions and AI usage data, then interview the CTO and engineers. Each finding is costed and routed to the price, the SPA, the deal structure or the 100-day plan.

What should be in the technology folder of a data room?

A technology overview, then folders for IP and ownership, open source, AI-generated code, security, data protection, architecture and operations, technical debt and roadmap, team, documentation, cloud cost, product metrics, the AI evidence pack, contracts and suppliers, internal IT, and vendor reports with the Q&A log. Every document should have existed before the process, with a named owner, and every claim in the information memorandum should point to one of them.

Is vendor technical due diligence worth it for a mid-market sale?

Usually, when technology is central to value and the process is competitive. One independent report replaces several sets of buyer questions, shortens the timetable and stops each bidder inventing its own estimate of the technical debt. It is worth less in a bilateral sale to a buyer who will run deep diligence anyway, or where technology is peripheral to value. A VDD report that exposes problems the seller then ignores is worse than none.

How do you prepare a CTO for technical due diligence interviews?

Build a fact base the CTO knows cold, run a mock due diligence six to eight weeks before launch with someone asking the questions a buyer's adviser will ask, and check every answer against the information memorandum and the vendor report. Agree who answers what, practise the live demo on real data, and settle the rules: answer the question asked, give numbers that match the data room, disclose known issues before being asked, and log anything you cannot answer for a written reply.

Take it with you
Free to use, adapt and share with attribution.
Brannan Coady, founder of Anyway Ventures
Written by Brannan Coady
Founder of Anyway Ventures. Co-founder and CEO of Netsells for thirteen years, then Chief Product Officer, COO and CEO of YourParkingSpace and a member of the Flowbird Group executive committee through its sale to EasyPark (now Arrive). Investor and board member; now leads technology and AI due diligence and delivery for PE-backed and mid-market businesses. More about Brannan →

Related

ARTICLE · 10 MIN
Vendor technical due diligence: what to fix before a sale
ARTICLE · 9 MIN
Will buyers pay for your AI? Building an AI story for exit
ARTICLE · 9 MIN
What a technical due diligence report should contain
RESOURCE · PDF + XLSX
The state of UK parking technology 2026
RESOURCE · PDF + XLSX
Technical due diligence checklist
RESOURCE · PDF + DOCX + XLSX
AI due diligence question bank

Want a second pair of eyes on it?

The first conversation is free and takes under an hour. If there's an engagement worth having you'll leave knowing its shape and its price. If there isn't, I'll say so.

Book 30 minutes brannan@anywayventures.com