Marketing measurement diligence is the pre-close test of whether a target's measurement setup was capable of producing the marketing numbers in the buyer's model. It is not a review of whether those numbers look reasonable. It is an examination of the instrument that produced them: what the setup counted, what it defined as a conversion, how much of the company's real demand it never saw, and how much of that can still be corrected after the deal closes.
The distinction matters because marketing numbers fail differently from financial ones. A ledger can be reconciled. A pipeline export can be recounted, which is the subject of a separate test list for a target's CRM and pipeline data. Measurement is different. When a tag was missing for nine months, the demand it would have recorded is gone. No post-close project recovers it. That makes the pre-close question narrower and more urgent than it first appears, and it is a different question from which numbers to request in the first place, which is covered in the deal team's marketing KPI ask list.
A measurement audit that starts as a checklist produces a list of configuration faults and no judgment. Start instead with a single test that produces a verdict.
Pick one channel and one closed month from the figures already in the model. Ask for the raw export behind that figure rather than the dashboard or the screenshot: the event or session level file, the ad platform export for the same dates, and whatever file the company uses to tie either of them to revenue. Then rebuild the number.
Three outcomes are possible.
The third outcome changes how much weight the rest of diligence can carry. A number nobody can reproduce is an assertion, and assertions should not be underwritten at the same confidence as measurements.
Every measurement defect found in diligence falls into one of two groups, and the split decides what a buyer can do about it.
Definitional failures. The event was captured, but it counts something other than what the model assumes. A conversion that is a form submission counted as a customer. A quote request counted as a sale. A thank-you page that fires twice on a refresh. A booking counted at request rather than at completion, in a business where a third of requests never turn into work. These are recoverable. The raw events exist, so the history can be restated once the definition is corrected.
Coverage failures. The demand happened and nothing recorded it. A tag absent from the booking subdomain. A consent banner that blocks measurement for a share of visitors who then buy. A phone line that never carried a source. A location whose walk-in work never touches a digital system. These cannot be fixed backwards. There is no event to recount, so every month in the trailing period is permanently understated by an unknown amount.
Buyers routinely treat both groups as a single remediation line in the hundred day plan. They are not comparable. A definitional failure costs analyst time and changes the numbers on the page. A coverage failure changes what the buyer is able to know, for the whole period the plan will be measured against.
Each of these is answerable in a short working session with whoever owns the reporting, and each has a pass condition rather than a rating.
A target that fails several of these is not necessarily badly run. Definitional drift is what happens when a measurement setup is built in stages by different people over several years, which describes most mid-market companies.
Coverage is harder to test because the failure leaves no trace inside the system that failed. Nothing in a platform tells a buyer what the platform never saw.
The only reliable way to measure coverage is from outside the measurement stack. Take total customers or jobs for a period from the finance system, where every one of them appears because every one of them was invoiced. Then take the number of those the measurement setup attributed to any source at all. The residual is the blind spot, and it is a measured figure rather than an estimate.
That residual is usually larger than the company expects, and it is not noise. In Claymore's own work, one engagement found that 38 percent of a client's new business arrived through word of mouth that no system had been recording. A buyer who found the same residual pre-close would be looking at an acquisition cost figure computed on under two thirds of the actual customer base, and at a growth plan built on channels that were never the main source of demand.
A large unattributed share in a business with strong repeat and referral revenue is a different finding from the same share in a business whose plan depends on paid acquisition. The first says the company is healthier than its reporting and the paid channels are smaller than they look. The second says the plan is being measured against an instrument that cannot see most of what it is supposed to manage.
Measurement assets travel badly, and the question is easy to leave until it is too late to answer well.
None of this changes the price on its own. All of it changes the sequence of the first hundred days, and it is cheaper to secure an account before signing than to reconstruct a year of history afterward.
Findings earn their place in the deal conversation by naming the number they move. Measurement findings do one of three things, and the three are not interchangeable.
They change a number. A definitional failure that counted quote requests as customers makes the reported acquisition cost too low and the reported conversion rate too high. The corrected figure is computable, and a figure that moves belongs in the price conversation.
They widen the band around a number. A coverage gap of unknown size means the acquisition cost is somewhere in a range rather than at a point. That is a plan conversation, carrying a remediation cost and a date, not a price adjustment, because the buyer cannot yet say which direction the true figure sits in.
They remove the ability to measure the plan. This is the case buyers skip. If the growth plan depends on shifting mix toward owned demand, and the setup cannot distinguish owned demand from paid demand, then the plan has no instrument. The cost is not a haircut on entry. It is a year of the hold spent unable to tell whether the thesis is working, with every subsequent spending decision made on the same bad information.
That third category is also the argument for doing this work pre-close rather than treating it as a post-close project. Remediation after close can repair the instrument going forward. It cannot manufacture the trailing history the plan will be measured against.
This is one test list inside a larger review. The surrounding workstreams, what each one covers and what each one does to price are set out in the guide to digital due diligence scope, and the commercial review this sits within is described in the guide to marketing due diligence. The measurement tests here are the ones that decide how much confidence the rest of that work can carry, because every channel, cost and payback figure in a commercial model was produced by the setup being tested.
Claymore runs this as part of a Commercial Audit First approach, on the principle that a buyer should establish what can be measured before underwriting what will be grown.
Marketing measurement due diligence is the pre-close test of whether a target's measurement setup was capable of producing the marketing numbers the buyer has been given. It does not review the numbers for plausibility. It asks whether a conversion in the reported figures means what the model assumes it means, how much of the target's real demand the setup never recorded, and which reported figures can still be restated after close. The work sits inside digital and marketing due diligence rather than replacing either one.
Run one reproduction test before anything else. Take a single channel figure for a single closed month, ask for the raw export behind it rather than the report, and rebuild the number. Three outcomes matter. The number reproduces, and the rest of the test list is about coverage. The number reproduces only with an undisclosed filter or date setting, which is a definitional problem. Or it cannot be rebuilt at all, which means nobody at the target can currently show where the figure came from.
A definitional failure means the recorded number is real but counts something other than what the buyer assumes, such as a form submission counted as a customer. Those failures can be restated later, because the underlying events were captured. A coverage failure means the demand happened and was never recorded at all, through missing tags, consent refusals, phone calls or offline bookings. Coverage failures cannot be recovered backwards, because there is no event to recount.
Not reliably. Analytics properties, ad accounts and tag containers are often owned by an agency, a former employee or the seller rather than the company, and a property that has to be rebuilt starts with no history. Data retention settings also delete event-level detail on a rolling basis regardless of ownership. A buyer should establish who holds each account, what gets transferred at close, and whether the first post-close year will have a comparable prior year to measure against.
Only the ones that move a number in the model. A gap that shows a reported acquisition cost was understated changes the figure being underwritten and belongs in the price conversation. A gap that only widens the uncertainty around a number belongs in the plan, with a remediation cost attached. The expensive third case is a gap that leaves the buyer unable to tell after close whether the growth plan is working, which is rarely priced and should be.
Have a revenue problem the board is asking about? Start a conversation.