The Anatomy of a Credible Digital Case Study

Case studies are the easiest sales asset to polish. They report outcomes, and outcomes can always be selected: the right period, the most flattering metric, the lowest baseline. Reading a case study without an evaluation framework is therefore like judging a restaurant from photographs of its menu.

This article dissects the anatomy of a digital case study: which parts must be present for a claim to be assessable, how polished numbers give themselves away, and the questions that leave claims nowhere to hide. It is useful in both directions — for readers assessing a potential partner, and for anyone writing their own case study who wants it to survive scrutiny.

The six mandatory parts

A case study worth assessing contains six parts. Missing even one makes the rest hard to verify.

1. Starting context. Business type, size, market, and conditions before the work began. Without it, any number loses meaning.
2. The defined problem. Not "wanted more revenue" but a specific, measurable problem.
3. Actions taken. Detailed enough that a reader could imagine repeating them.
4. Time frame. When it started, when it was measured, how long it ran.
5. Results with baselines. Not only "up 300%" but from what to what.
6. Limits and other factors. What else happened in that period that may have contributed.

The sixth part appears least often and matters most for credibility. A case study stating "during the same period the client also added two salespeople" is far more trustworthy than one presenting results as though they flowed from a single action.

How numbers get polished without lying

Most misleading numbers in case studies are not outright lies. They are favourable selections. Recognising the patterns lets you ask precisely the right question.

A very small base

Going from two leads to eight is a 300% increase. Technically true, commercially close to meaningless. Always ask for the starting number — large percentages from small bases are the most common tell.

A chosen period

Comparing a peak month against a quiet one produces a rise that is really seasonality. Ask whether the comparison is the previous period or the same period last year — the second is far more honest for cyclical businesses.

Shifting metrics

A claim opening with "increased traffic" and closing with "increased sales", without connecting the two, leaves an unproven leap. Watch whether the metric promised at the start is the metric reported at the end.

Results without costs

Doubling lead volume is easy if you triple the ad budget. A case study quoting results without quoting spend removes half the information needed to judge it.

Relative numbers without market context

If the whole industry grew forty percent in that period, a fifty percent rise is not a triumph. Market context is rarely mentioned even though it changes the conclusion.

Carefully chosen numbers are not lies. But a case study that never says what was left out is still misleading.

Twelve questions claims cannot hide from

Ask these when assessing a potential partner. A nervous or circling answer is usually more informative than the answer itself.

• What was the starting number, not the percentage?
• How long was the measurement period?
• What was it compared against — the previous period or the same period last year?
• What did it cost to achieve that result?
• What else changed in the client's business during the same period?
• How was the result measured, and who measured it?
• Did the result hold after the work ended?
• How many similar projects did not reach this result?
• Which part of the work was hardest, and what failed on the first attempt?
• Can we speak to the client?
• Under what conditions would this approach not work?
• What would you do differently if you repeated it today?

The final three are the most revealing. A partner who answers them specifically almost certainly did the work; one who cannot is usually reciting marketing material.

Writing your own case study

If you are the author, the same framework applies in reverse: cover the six parts, and state what is normally hidden. The effect is counterintuitive — acknowledging limits raises trust rather than lowering it.

Writing without exposing client data

Many clients will not allow specific numbers to be published. That is not a licence to invent, nor a reason to write nothing. Several approaches remain honest and useful:

• Ask for written permission for the figures you want to quote. Partial permission is often granted.
• Use ranges instead of exact figures — "from dozens of leads a month to hundreds" is still informative.
• Anonymise the identity but name the industry and business size, since that determines relevance.
• Focus on process rather than outcome. Describing how the problem was diagnosed is often more convincing than the final number.

What not to do: combining figures from several clients and presenting them as one case, and presenting results as a guarantee that will repeat. Both are common and both mislead.

Customer statements and social proof

Besides numbers, case studies usually carry statements from customers. This section is often written in a way that reduces trust: generic praise that could be pasted onto any company.

Convincing statements share three traits. First, they describe a specific situation rather than a general quality — "we had no idea where our leads came from before this" is stronger than "excellent service". Second, they mention an initial doubt, because someone who was once sceptical is more believable than someone enthusiastic from the start. Third, they come from a person with a name, role, and business that can be checked.

If a client will not be named, briefly say why. "A financial sector client who cannot be named under internal policy" is far better than an unexplained anonymous quote, which always invites the suspicion that the client does not exist.

Signs a statement should not be trusted

• Several statements written in an identical voice — usually one author.
• Profile photos that do not match the name or look like stock imagery.
• Numbers quoted inside the statement that never appear in the results section.
• Dates all clustered together, indicating they were collected in one marketing push.

Separating cause from coincidence

Causal claims are the most fragile part of any case study. Sales rising after a new site launched does not automatically mean the site caused it — there may have been another campaign, a season, or a competitor closing down.

Three ways to strengthen a causal claim without academic research:

1. Change one thing at a time. If site, ads, and pricing all changed in the same month, nothing can be concluded.
2. Use a comparison. An unchanged page, an uncampaigned region, or an untouched product acts as a simple control group.
3. Check the timing. If the rise started two weeks before the work finished, the work was not the cause.

These principles apply just as much to assessing your own work as to assessing someone else's claims. The measurement framework is in our marketing ROI guide.

Whether results survive the engagement

A question case studies almost never answer: what happened afterwards. Many digital results are temporary because they depend on ad spend that keeps flowing, or on one person handling things intensively who then leaves.

Distinguishing durable results from temporary ones takes one simple question: if the engagement stopped today, what remains? The answer reveals whether what was built is an asset or a flow.

• Assets keep working after the work stops: pages that rank, better site structure, a lead recording system, a customer list, or a process the team now runs itself.
• Flows stop when payment stops: paid traffic, daily-managed campaigns, or a spike from one promotion.

Both are legitimate and often needed together. What to watch for is a case study selling a flow in the language of an asset — implying results will persist when in fact they end with the budget.

A useful follow-up question

"What were the results six months after the engagement ended?" is a question nobody prepares an answer for, and that is precisely why the answer is informative. A partner who tracks clients after projects end usually works differently from one who never looks back.

Judging relevance, not only truth

A case study can be entirely honest and still irrelevant to you. This is a subtler assessment error than being fooled by numbers, and it happens more often.

Four axes of relevance to check. First, business model similarity: results from an online store with thousands of small transactions barely transfer to a project services firm with five large clients a year. Second, size: tactics that work on a large budget often have no small version. Third, starting point: a business rising from zero has different headroom than an established one. Fourth, timing: an approach that worked three years ago may no longer apply as ad costs and search behaviour have shifted.

Checking it takes one calmly asked question: "what makes you confident this approach transfers to our situation?" A good answer names both the similarities and the differences, then explains which parts need adapting. A poor answer repeats that the method is proven.

Signs of a case study worth trusting

Good sign	Worth probing
States starting and ending numbers	Percentages only
States cost alongside result	Results with no cost
Names other contributing factors	Result attributed to one action
Says what failed	Everything went smoothly
Client is contactable	Anonymous client with no reason given
Says when the approach does not fit	Works for every business

When there are no case studies at all

Young firms often have none, and that alone is not grounds for rejection. What can substitute: a detailed process explanation, real output samples from previous work, clarity on how progress will be reported, and willingness to agree benchmarks up front.

In fact, willingness to agree benchmarks before work begins is a stronger signal than any case study. It shows the result will be measured rather than narrated afterwards.

Reading the proposal that accompanies the case study

Case studies rarely stand alone; they usually arrive with a proposal. A complete assessment therefore also checks whether the promises in the proposal are consistent with the story in the case study.

Three inconsistencies show up most often. First, the case study describes work spanning a year while the proposal promises comparable results in three months. Second, the case study mentions a large team while the proposal offers a price that only makes sense for one part-time person. Third, the case study relies on deep data access while the proposal mentions no access at all that you would need to grant.

None of these prove bad intent. Often the proposal is written by sales and the case study by delivery, and the two are never read side by side. But the inconsistency still needs raising, because what you will receive is the contents of the proposal, not the contents of the case study.

Agreeing benchmarks up front

The best way to close the gap between story and reality is agreeing benchmarks before work starts. Three things suffice: a starting number both sides accept, the metric that will be used to judge, and when the judgement happens. Written as one paragraph in a contract annex, this agreement saves months of later argument.

A partner who refuses to set benchmarks on the grounds that "there are too many variables" is telling you something important. Variables always exist; that is the reason benchmarks need agreeing, not a reason to have none.

In summary

A case study is assessable when it contains context, problem, actions, timing, results with baselines, and limits. Watch for small bases, chosen periods, shifting metrics, and results without costs. Ask questions demanding specifics, especially about what failed and when the approach does not fit. And if you are the author, state the limits — that is what makes the rest believable.

One closing thought for anyone writing their own: the most persuasive case study is rarely the one with the biggest numbers. It is the one a sceptical reader finishes without finding anything to argue with.

Want help assessing a proposal or writing an honest case study? Talk to our team.