Operated function · evidence
Willingness to say something good peaks the week the thing works and decays steadily after. By the time a case study is on a content calendar, the result has become normal, the person who cared has changed role, and the answer is a polite maybe.
The results exist. They are in a support thread, a delivery report, a renewal conversation, a number in a dashboard somebody screenshot once. What does not exist is any of it in a form that can be shown to a prospect, because turning a result into evidence requires asking somebody, and asking is nobody’s scheduled work.
When it is scheduled, it is scheduled by marketing on a content calendar, which is disconnected from when anything actually happened. The request arrives months late and reads as a favour rather than as a natural moment.
The approval chain is what kills most attempts. A customer contact says yes enthusiastically and then discovers they need their communications team, their legal team, and sometimes a procurement clause checked — and each of those is a delay in a process that had no urgency to begin with.
Meanwhile every salesperson builds their own evidence. They assemble something from memory for each deal, describing outcomes they half-remember from accounts they were not on, which is both weaker than a documented account and a claim exposure nobody is tracking.
And the evidence that would matter most is the specific kind: a named organisation, a stated starting position, a measured change. What usually exists instead is a quotation saying the team was helpful, which persuades nobody who was going to ask a hard question.
A customer’s willingness to provide evidence is highest immediately after a result and falls steadily afterwards, while the request is triggered by a marketing calendar — so almost every ask happens well past the point of easiest yes.
Fixing the timing is most of it. If the ask is triggered by the result rather than by a calendar, it arrives while the person still remembers being pleased, and the conversation is about something specific that just happened rather than about a favour for a marketing programme.
Which means the results have to be observable. That is the unglamorous prerequisite: knowing when something good happened, from your own systems, rather than from somebody mentioning it. Most businesses can do this and have never wired it, because nobody owned the question.
The approval chain has to be worked as a process rather than discovered as an obstacle. Knowing at the outset that this customer requires communications sign-off, and starting that in parallel rather than after the contact agrees, is the difference between a case study in six weeks and one in nine months.
And what gets captured should be the specific thing: the starting position, what changed, over what period, measured how. A quotation is easy to get and persuades nobody who matters; a measured change is harder and is the only thing a sceptical buyer will accept.
What is never delegated is the claim. What your organisation asserts publicly about a customer’s result is a marketing and legal decision, and every asset stays unusable until your people approve it.
Time from result to ask — measured by days between the observed result and the first request, against a baseline where the trigger was a content calendar.
Documented, approved evidence assets — measured by the count usable in a deal, against a starting point that is frequently zero despite many good results.
Evidence that states a measured change — measured by assets carrying a starting position, a change and a measurement, versus quotations about helpfulness.
Time through the approval chain — measured by days from customer agreement to usable asset, before and after approvals are worked in parallel.
Claims made in deals that nobody approved — measured by the count of salesperson-assembled assertions, which is an exposure most organisations have never quantified.
Customers asked more than once by different teams — measured by duplicate requests, which is the fastest way to exhaust goodwill you were relying on.
that customers will agree, and no approval of any public claim. A meaningful share will decline for reasons that have nothing to do with satisfaction — policy, competitive sensitivity, a communications team that says no to everything. Nothing here approves what your organisation asserts publicly; that stays with your marketing and legal teams.
Result observation reads your own product and delivery data, because the trigger has to be something that happened rather than something somebody mentioned. Where a result genuinely cannot be observed, the trigger falls back to a human flag and that limitation is stated rather than papered over.
Approved assets land where your sales team already looks. An evidence library nobody can find during a deal is why salespeople assemble their own, which is the exposure this operation exists to reduce.
The approval chain is tracked per account, because it differs per account and because knowing it at the outset is what shortens everything.
Nothing is offered in exchange. An incentivised endorsement is a different object from a voluntary one, it has to be disclosed in most jurisdictions, and it is worth less to the sceptical reader it is aimed at. Where your organisation chooses to incentivise, that is a marketing decision with a disclosure obligation attached and it is made by your people, not defaulted into by an operation.
Nothing is published until your marketing and legal have approved it, and nothing is published that the customer has not seen in the form it will appear. A customer discovering their words in an unfamiliar context is a relationship problem and sometimes a contractual one.
Frequency is capped by your rule. The same customer asked repeatedly across teams exhausts exactly the goodwill the operation depends on.
Operational access is not permission to train. Customer result data — which describes their business performance — does not become material improving anything serving another organisation, including their competitors.
Legal owns the claim standard: what may be asserted about a customer result, what evidence supports it, and what disclosure is required if anything was given in exchange. That standard should exist before the first ask, because it determines what is worth capturing.
Account owners must agree the ask, because it goes to customers they are responsible for and an ask that lands during a difficult period costs them a relationship.
Where an obligation attaches through endorsement disclosure rules, your customer contracts, or a sector restriction on naming clients, it is marked applicability-gated rather than presented as standing.
One trailing period — read-only, nobody contacted — identifying customers whose observable results would have justified an ask, and how long ago each one occurred.
The first phase asks nobody. It looks at your own data for results that happened and were never converted into anything, and it puts a date on each. That list is usually longer than expected and the dates are usually older than anybody would like.
It also establishes the approval chains for those accounts, which is the part that determines feasibility. An account whose communications team refuses all vendor references is not a candidate regardless of how good the result was, and knowing that early saves the ask.
If you continue, the first delegation is the ask itself on newly-observed results only — while they are fresh — with the claim standard written and approvals worked in parallel from the outset.
Some sectors genuinely are like that, and the first phase establishes it per account rather than assuming it either way — the approval chains are mapped before anybody is asked. Where naming is impossible, anonymised measured results are frequently permitted and are worth considerably more than a named quotation about helpfulness. Where even that is refused across the board, this is the wrong operation and the honest answer is that your evidence has to come from somewhere else.
Then the question is what triggers it. If the trigger is a content calendar, the asks are arriving months after the results and the yield reflects that. If it is a result observed in your own systems within days, this is a duplicate. The first phase measures the gap directly — result date against ask date — using data you already hold.
Repeated, mistimed, generic asks do. An ask that arrives days after something specifically went well, about that specific thing, from the business it went well with, is a different interaction — and the frequency cap across teams exists precisely because the damage comes from being asked four times by four people. Where an account is in a difficult period, it is excluded by rule rather than by somebody remembering.
You can, and it changes what the thing is. An incentivised endorsement carries a disclosure obligation in most jurisdictions and is worth less to the sceptical buyer it was collected for, because they will assume it was bought. That is a marketing decision your organisation may legitimately make — it is not defaulted into here, and if it is made, the disclosure is part of the asset rather than an afterthought.