Most consulting deliverables are never audited. The report arrives, the steering group nods, the invoice is paid, and eighteen months later someone asks what the £400,000 bought and nobody can say. The UK’s Public Accounts Committee asked exactly that about a programme that at one point had 2,239 consultants on it at an average of £1,100 a day, and did not get a good answer.
You do not need a parliamentary committee. You need five tests, a rubric, and the discipline to run them at three points: in the middle of the engagement, at handover, and a year later.
Auditing consulting deliverables
Why audit at all
Three reasons, in increasing order of importance.
To decide whether to pay the final invoice. Most consulting contracts stage payment, and the last stage is due on acceptance of the deliverables. Acceptance should mean something. If the deliverable fails the tests below, you have a basis for rework before you pay.
To decide whether to act on it. A recommendation you have not checked is a recommendation you are taking on trust. The tests are the check.
To decide whether to hire them again. Consultants get rehired on relationships, and relationships are a poor guide to whether the last engagement delivered. An audit gives you something better than the memory of a good final presentation.
The five tests
Test 1: Traceability to the brief
Lay the original brief, or the proposal’s statement of scope, next to the deliverable. Every question in the brief should have an answer in the deliverable, and every major section of the deliverable should trace back to a question in the brief.
Two failure modes. Drift: the brief asked why margins were falling and the report is about digital transformation. Substitution: the brief asked for a recommendation and the report offers “options for the executive team to consider”. Both are common; both mean the consultant answered a question they preferred to the one you asked.
Pass: each question in the brief is answered, explicitly, with the answer stated in a sentence rather than implied across a section.
Test 2: Evidence you can check
For every material finding, ask: what is this based on, and can I see it? Interviews should have a list of who was interviewed. Data analysis should come with the data and the model. Benchmarks should name the comparator companies or the published source, with a date.
This is where most deliverables fail, and where the failure is easiest to hide, because a well-designed slide looks like evidence. The test is not whether the slide is convincing; it is whether the number on it can be traced to something outside the consultant’s laptop. We hold our own benchmarks to this: every figure in our fees by industry guide has a named source and a verification date, and a consultant charging you $300 an hour can meet the same standard.
Pass: the underlying analysis, data and sources are handed over with the report, not just described in it.
Test 3: Originality
Would this report be different if it had been written for your nearest competitor? Some of it should not be: frameworks and structures are reusable, and that reuse is part of what you are paying for. But the findings, the numbers and the recommendation should be specific to your organisation.
The quick check is to search the deliverable for your company’s name, your products and your people. If they appear only on the title page and in a few section headings, the body is a template. The slower check is to ask the consultant which three findings surprised them. A consultant who has actually done the work will have an answer.
Pass: at least half the material findings could only have been written about you.
Test 4: Actionability
A recommendation is actionable if it says what to do, who should do it, by when, what it costs and what it is expected to produce. “Strengthen commercial capability” is not a recommendation. “Hire a head of pricing by Q2, at £120,000 to £140,000, reporting to the CFO, with a target of 2 points of gross margin by year end” is.
Count the recommendations and score each one against those five elements. The proportion that scores five out of five is the actionability rate. Below half, the report is a diagnosis without a treatment plan, and you should ask for the plan before you pay for the diagnosis.
Pass: every material recommendation has an owner, a date, a cost and an expected result.
Test 5: Ownership and handover
At the end of the engagement, what do you own? The report, obviously. But also the models, the data, the interview notes, the workshop outputs and the working files. If the consultant proposes to keep the model “because it contains our proprietary methodology”, you have bought a picture of an answer rather than the answer.
Also under this test: can your team explain the deliverable without the consultant in the room? If the only person who can present the findings to the board is the consultant, the knowledge has not been transferred, and the next question the board asks will cost you another engagement.
Pass: you hold the working files, and someone on your staff can present and defend the findings unaided.
Score each test as 0 (fail), 1 (minimum) or 2 (upgraded). A deliverable scoring below 5 out of 10 should not be accepted without rework. Below 3, escalate to the partner.
When to run the audit
Mid-engagement. Run tests 1 and 2 against the interim findings at the halfway point. This is the audit that changes outcomes, because the consultant can still redirect the work. If there are no interim findings to audit by the halfway point, that is itself a finding, and the reason we say in our guide to what consultants actually do that you should see evidence from week two.
At handover. Run all five tests before accepting the deliverables and releasing the final payment. Give the consultant the rubric in advance; a good one will welcome it, because it tells them exactly what acceptance means.
Twelve months on. Run the test that matters: what happened? Which recommendations were implemented, what did they produce, and how does that compare with what the report promised? This is the audit that tells you whether the engagement was worth the money, and it is the one almost nobody does.
Was it worth the money?
The twelve-month audit reduces to arithmetic, provided you set it up at the start.
Before the engagement: write down the fee, the expected result (the number the consultant’s recommendation is supposed to move), and how you will measure it.
Twelve months after: compare the measured result with the expected one, and the value of the result with the fee. A £150,000 engagement that produced a recommendation which, when implemented, added £600,000 of annual margin was worth it four times over. One whose recommendations were not implemented was worth nothing, and the useful question is why: was the advice wrong, or was the organisation unwilling? Those have different remedies.
First, nobody wrote down the expected result at the start, so at twelve months there is nothing to compare against and the engagement is judged on whether people liked the consultant. Second, the recommendation was implemented, the result was achieved, and the consultant claims all of it, when most of the work was done by your staff. Write down the expected result and who does what before the engagement starts, and both problems disappear.
For engagements on outcome-based or value-based fees, still a small share of the market (3.3% of management consulting work in SPI Research’s 2025 benchmark), this audit is not optional. It is the invoice. Our consulting pricing guide covers how those fee structures are set up so that the measurement is agreed in advance.
Was the engagement worth it? Five questions
Question 1 of 5
Were the brief's questions answered with evidence you could check?
What share of the recommendations were implemented within twelve months?
Did the implemented recommendations produce the result the report predicted?
Can your team explain and defend the findings without the consultant?
Would you have reached the same conclusion without them?
When the deliverable fails
The moment to raise a failed deliverable is before the final payment, in writing, against the rubric you gave them at the start. Raise it afterwards and you are asking a favour. Raise it before and you are enforcing a contract.
Rework before payment. Most contracts allow acceptance testing. Send the rubric scores, name the failing tests, and ask for rework within a defined period. Reasonable consultants will do it; the partner’s reputation is worth more than the final invoice.
Withhold the final stage. If rework is refused, the staged payment structure exists for this. Withhold the acceptance-linked payment and cite the contract’s acceptance clause. This is why the guide to hiring a consultant says to stage payments and define acceptance before signing.
Escalate to the partner. Engagement managers protect their teams; partners protect the account. A partner who learns that a deliverable failed the client’s rubric will usually fix it.
Record it. Whether or not the dispute is resolved, write down what happened. Firms are rehired on memory, and memory favours the final presentation.
- Audit consulting deliverables with five tests: traceability to the brief, checkable evidence, originality, actionability, and ownership of the working files
- Score each test 0 to 2; do not accept a deliverable below 5 out of 10, and escalate to the partner below 3
- Run tests 1 and 2 at the halfway point, all five at handover before the final payment, and the results audit twelve months later
- Write down the fee, the expected result and the measurement method before the engagement starts, or the twelve-month audit has nothing to compare
- An unimplemented recommendation was worth nothing; the useful question is whether the advice was wrong or the organisation was unwilling
- Raise a failed deliverable before the final payment, in writing, against a rubric the consultant saw at the start
Frequently asked questions
How do you evaluate a consultant’s deliverables?
Against five tests: does the deliverable answer every question in the brief; is every material finding traceable to evidence you can check; is the content specific to your organisation rather than a template; does every recommendation have an owner, date, cost and expected result; and do you own the working files and understand the findings well enough to present them yourself. Score each 0 to 2 and set an acceptance threshold before the engagement starts.
When should you audit a consulting engagement?
Three times: at the halfway point against interim findings, when the work can still be redirected; at handover, before accepting deliverables and releasing final payment; and twelve months later, to compare what was implemented and what it produced with what the report promised.
How do you know if a consulting engagement was worth the money?
Write down the fee, the expected result and how you will measure it before the work starts. Twelve months after, compare the measured result with the expected one and its value with the fee. If the recommendations were not implemented, the engagement was worth nothing, and you need to know whether that was the advice or the organisation.
What if a consultant’s report is just a template?
It fails test 3. Search the document for your company’s name, products and people; if they appear only in headings, the body is generic. Ask the consultant which findings surprised them and for the evidence behind each. Request rework against the rubric before paying the final stage.
Can you withhold payment for poor consulting work?
If the contract stages payment against acceptance of deliverables, yes: withhold the acceptance-linked stage, cite the acceptance clause and the rubric, and request rework. If the contract has no acceptance clause, you are negotiating rather than enforcing, which is why the clause should be there before you sign.
Last updated: 28 September 2026