← ALL WRITING
DILIGENCE

What a technical due diligence report should contain

A section-by-section walkthrough of a tech due diligence report that a deal team can act on, with a sample findings table and the padding to cut.

WRITTEN
READINGNine minutes
FORDeal teams, operating partners, investment committees, deal lawyers

A technical due diligence report sets out what a buyer or investor is getting in a target's technology: architecture, code, delivery, team, security, IP, running costs and, increasingly, AI. A good one opens with a one-page verdict, then a register of findings in which every material issue is costed in pounds and routed to the price, the SPA, the 100-day plan or the file. Everything else is supporting evidence, and most of it belongs in an appendix.

That last sentence is where most reports fail. I've read tech DD reports that ran past a hundred pages and still didn't say whether the deal should proceed at the agreed price. I've also been on the receiving end, as a founder and CEO being diligenced, and seen how little of a long report made it into the negotiation. The structure below is the one I use and the one I'd want to receive.

If a finding has no route, it is an observation, and observations belong in an appendix.

It covers the deliverable. The evidence to request, area by area, is in the technology due diligence checklist and the numbered request list in our technical due diligence checklist, and the questions to put to the team are in technical due diligence questions.

What's in a technical due diligence report?

A full report for a software or tech-enabled target should have these sections, in this order:

  1. Executive summary and verdict
  2. Scope, method and limitations
  3. Findings register
  4. Architecture and scalability
  5. Engineering delivery and code quality
  6. Team and key-person risk
  7. Security and data protection
  8. IP, open source and third-party dependencies
  9. Product and roadmap
  10. Cost to run and technical debt
  11. AI (where the target claims it or is exposed to it)
  12. 100-day plan
  13. Appendices

A red-flag report is a cut-down version: the executive summary, a short scope statement and the findings register, with nothing else unless a finding needs explaining.

Executive summary: the only page some readers will see

The executive summary should fit on one page and answer three questions. Should the deal proceed? At what adjustment to price or terms? What has to be true in the first 100 days?

Open with a verdict in one sentence: proceed, proceed with conditions, renegotiate, or don't proceed. Then list the top three to five findings, each in a line with its cost range and route. Then one paragraph on whether the technology can deliver the investment case, with the specific growth assumption tested ("the platform can carry the 2.5 times customer growth in the plan without re-architecture, but not the move into enterprise accounts in year two").

What it should not contain is a summary of every section. If the reader needs to turn to page 40 to find out whether there's a problem, the summary hasn't done its job.

Scope, method and limitations

This section looks dull. It is the one lenders and insurers read most carefully.

State what was reviewed (which repositories, at which commit, which environments, which documents), who was interviewed and for how long, and what was not reviewed and why. If the seller refused code access, or a codebase was out of scope, say so plainly.

This matters for warranty and indemnity insurance. Paul Hastings' September 2026 buyer's guide to W&I notes that the scope of protection depends heavily on the quality of diligence, that underwriters hold a call to discuss diligence scope and findings, and that cyber cover depends on a review of the target's IT infrastructure and security. WTW's guidance on exclusions makes the same point more bluntly: cover starts from the areas that were diligenced, and deal-specific exclusions tend to appear where diligence was thin. A vague scope section invites a broad exclusion. A precise one gives the underwriter less reason to carve out technology.

The section should conclude with a clear reliance statement: who may rely on the report, and on what basis.

The findings register: how findings should be rated and costed

This is the heart of the report, and where good and bad reports differ most.

Most reports rate findings on a red, amber and green scale, or on severity times likelihood. That's fine as far as it goes, but it doesn't tell the deal team what to do. I'd rate every finding on three things.

Severity, in plain terms: critical (could stop the deal or cause material loss), high (should change price or terms), medium (should be in the 100-day plan), low (note for the file).

Cost, as a range in pounds with the basis stated. Engineer-months at a stated loaded rate, a supplier quote, a licence fee, a revenue figure at risk. A range is honest. A single number implies precision nobody has. If something genuinely cannot be costed yet, say what it would take to cost it.

Route, meaning where the finding goes:

  • a price adjustment or a debt-like item in the model
  • a condition precedent, something that must be fixed before completion
  • a specific warranty or indemnity in the SPA
  • a structural protection such as retention, deferred consideration or an earn-out term
  • the 100-day plan, with an owner and a budget
  • the file, for information only

If a finding has no route, it is an observation, and observations belong in an appendix.

Sample findings register (illustrative)

The table below is illustrative. The target, findings and figures are invented to show the format and are not drawn from any real deal. The costs assume a loaded engineering cost of £85,000 a year.

# Finding Severity Cost estimate Basis Route
1 Core billing service runs on a framework version out of security support since 2024; upgrade blocks the planned enterprise tier High £180k to £260k, six to nine months Three engineers for six to nine months, plus contractor cover Price adjustment as known capex; 100-day plan
2 71% of commits to the core platform in the last 12 months by the CTO, who holds sole admin access to the production cloud account and domain registrar High £90k to £150k to mitigate Senior hire, six months' overlap, documentation sprint Retention package; access transfer as a condition of completion
3 Early contractor who wrote the first version has no signed IP assignment High £2k to £10k to regularise if the contractor cooperates Legal cost of a confirmatory assignment Condition of completion; specific indemnity
4 AGPL-licensed library inside the hosted product Medium £30k to £50k to replace Two engineers for two to three months Specific indemnity; 100-day plan
5 Two high-severity pen test findings from March still open Medium £8k to £15k Engineering time, plus retest Condition of completion or pre-completion undertaking
6 Cloud cost up 48% year on year against revenue up 22%; no cost allocation by customer Medium £60k to £90k a year in avoidable spend Rightsizing and reserved instances, from bill analysis 100-day plan; adjust run-rate EBITDA
7 "AI-powered" matching in the CIM is a rules engine with one optional third-party model call, used by 6% of customers Medium No direct cost; revenue case for the AI tier discounted Usage data Model adjustment; disclosure
8 No staging environment; releases are manual and fortnightly Low £20k to £35k One engineer for three to five months 100-day plan

The arithmetic in finding 1, for example, is three engineers × £85k × 0.5 to 0.75 of a year, which gives £128k to £191k, plus contractor cover for the features those engineers would otherwise have built. Show your working like that. A deal team that can see the assumptions can argue with them, which is the point.

The technical sections, and what each should conclude

Each technical section should end with a conclusion of one or two sentences that a non-technical reader can repeat to the investment committee. The detail is there to justify that conclusion.

Architecture and scalability should conclude whether the platform can carry the volume and the product changes in the investment case, and, if not, what breaks first and what fixing it costs.

Engineering delivery and code quality should conclude how fast and how safely the team ships. DORA's software delivery metrics, which split into throughput (change lead time, deployment frequency, recovery time from a failed deployment) and instability (change fail rate, deployment rework rate), are the most widely used way of putting numbers on this. They are better evidence than a static code-quality score. A conclusion might read: "The team deploys several times a week with a low failure rate. Delivery speed will not constrain the plan."

Team and key-person risk should conclude whether the product survives the departure of any one person, and if not, who, and what it costs to fix. This is often the most valuable section on a smaller deal, and the method is in key person risk in due diligence.

Security and data protection should conclude whether there is evidence of an unreported incident, how the target handles personal data under UK GDPR, and what the open findings cost to close. A SOC 2 or ISO 27001 certificate is a signal of process, not a conclusion.

IP, open source and third-party dependencies should conclude whether the target owns what it sells, whether any licence obligation threatens the commercial model, and whether any single supplier could stop it trading.

Product and roadmap should conclude whether the revenue plan depends on features that don't exist yet, and how far to trust the team's delivery estimates, judged by last year's roadmap against what shipped.

Cost to run and technical debt should conclude with numbers: hosting and tooling as a share of revenue, the trend, and the total costed debt that belongs in the model.

AI, where relevant, should conclude whether the AI described in the CIM exists, what it costs per use, and whether AI threatens the revenue model. I cover the method in AI due diligence for private equity, and the questions behind it in our AI due diligence questions.

The 100-day plan

The report should end with the actions that follow from the findings, each with an owner, a budget and a deadline. Keep it to what the findings justify: secure admin access, close the open pen test items, hire the second senior engineer, start the framework upgrade. A generic transformation roadmap does not belong here. If the fund has a separate value creation plan, link the two, as I describe in AI due diligence and the value creation plan.

What's useless padding in a tech DD report

Most of the length in long reports comes from material that doesn't change any decision. Cut or move to appendices:

  • A glossary explaining what an API or the cloud is.
  • Generic best-practice descriptions of what good engineering looks like, with no reference to the target.
  • Raw tool output (dependency lists, static analysis dumps, screenshots of dashboards) in the main body.
  • Maturity scores on a one-to-five scale with no cost and no route attached.
  • Restatements of what management said in interviews, without saying whether the evidence supported it.
  • Findings so hedged they can't be acted on: "there may be some risk of scalability challenges in future".
  • A list of sixty low findings presented with the same weight as the three that matter.

The opposite failure is worth naming too. A report so short it has no scope section and no stated assumptions can't be relied on by a lender or insurer, however sharp its conclusions.

Red-flag report, full report and vendor report

The same structure flexes for three purposes. A red-flag report, delivered in about a week, is the executive summary and findings register only, with rough cost ranges, written to support a go or no-go decision. A full buy-side report is everything above. A vendor technical due diligence report, commissioned by the seller, has the same sections but is written for several unknown buyers, so it states facts and costs more neutrally and comes with reliance letters. I discuss when a seller should commission one in vendor technical due diligence.

Whichever version you commission, ask for a redacted example before you sign the engagement letter. If the sample report has no costed findings register, the one you get won't either.

Our technical due diligence reports follow this structure, with a red-flag readout in the first week and every finding costed and routed. The fees and timelines are set out in how much technical due diligence costs in the UK, and the technical due diligence page explains how we scope the work.

FREE RESOURCE · PDF + XLSX
The state of UK parking technology 2026
DOWNLOAD FREE →

Questions

How long should a technical due diligence report be?

For a lower mid-market software deal, the main body should be 15 to 30 pages, with a one-page executive summary and technical detail in appendices. Length is a poor proxy for quality. A report of 120 pages on a £15m deal usually means the reviewer did not decide what mattered. The test is whether an investment committee member can read the first two pages and know the verdict, the top findings and their cost.

What is the difference between a red-flag report and a full technical due diligence report?

A red-flag report lists only the issues that could stop the deal or change the price, each with a rough cost and a recommendation, usually in five to ten pages after a few days of work. A full report covers every material area, costs every finding with stated assumptions, routes each one to price, the SPA or the 100-day plan, and sets out its scope and limitations so lenders and insurers can rely on it.

Can a buyer rely on a vendor technical due diligence report?

Only if the report's author gives the buyer a reliance letter, which is normally negotiated as part of a vendor due diligence engagement and may carry a liability cap. Without reliance, the report is useful background but the buyer has no claim against the author if it is wrong. Many buyers also commission a short top-up review to test the vendor report's key findings.

Who reads a technical due diligence report?

The investment committee reads the executive summary. The deal lawyers use the findings register to draft warranties, indemnities and conditions. The financial model team picks up the costed items as price adjustments or debt-like items. Lenders and the W&I insurer read the scope and limitations section. The operating partner and incoming management use the 100-day plan. A good report is written so each reader can find their part quickly.

Should the target see the technical due diligence report?

Usually not the buyer's full report, which contains negotiating positions. It is common, and helpful, to share factual findings with the target's management before the report is final so they can correct errors. Sharing the costed findings that justify a price chip is a negotiation decision for the deal lead, not the reviewer.

Brannan Coady, founder of Anyway Ventures
Brannan Coady
Founder of Anyway Ventures. Exited founder, former CEO of YourParkingSpace, investor and board member; now building AI inside PE-backed and mid-market businesses. More →

Keep reading

ALL WRITING →
DILIGENCE · 10 MIN
Technology due diligence checklist for private equity deals
Every tech DD checklist covers the same ground. The useful ones say which answers should change the price, the structure or the first 100 days.
DILIGENCE · 10 MIN
AI due diligence for private equity: what to test before you sign
Most AI diligence reports check whether the target's AI is real and stop there. That is one job of three, and it is rarely the one that moves the price.
DILIGENCE · 10 MIN
AI due diligence is a value-creation plan, not a risk register
Most funds now ask 'what's the AI risk?' in diligence. The better question is what AI makes this business worth, how fast, and what the IC paper should say about it.

Talk about this piece.

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