Notes for owners · Branding and communication
How to read a case study and spot the ones that prove nothing
Before you trust a case study from an agency or supplier, check the starting position, what was changed, who measured the result and whether the customer can be named. A six-question test.
The GullySales team · Updated 17 Oct 2026 · 5 min read
On this page
A case study proves something only when it tells you where the customer started, what was changed, how the result was counted and by whom. If it gives a big percentage with none of those, treat it as an advertisement. Read every case study with six questions, and give a point for each one it answers. Four or more and it is worth a phone call. Three or fewer and it proves nothing about what would happen to you.
The six questions
1. Where did they start? A result means little without a before. "Enquiries up 300 per cent" from two a month and from two hundred a month are different stories. Look for the starting figure, or at least a description: how many people, what channels, what was broken.
2. Is the customer like you? A hospital chain's result tells a corner workshop very little. Check the trade, size, town and kind of buyer. Better a modest case from your own situation than a dazzling one from someone else's.
3. What exactly was changed? The study should name the work: a new quotation format, a rebuilt Maps listing, a daily follow-up call. "A complete digital strategy" is not a thing anyone did.
4. Who counted the result, and from what? Enquiries from the customer's own sheet are stronger than clicks from the agency's dashboard. A figure that only the agency can see is a claim.
5. Can the customer be named and asked? A named customer who will take your call is strong proof. An anonymous "leading manufacturer" is weaker, and it is not worthless if the other answers hold.
6. Does it say what did not work? Real projects have a stalled month, a channel dropped, a mistake corrected. A case study with no friction may have been edited by someone who feared the reader.
A scorecard you can copy
Print the six questions with a tick box each and score every case study you are shown, one sheet per study. Mark a half point where a question is partly answered, such as a starting figure with no date range. Put the total and one line of doubt at the bottom. The cut-offs of four or more and three or fewer are our rule of thumb, not a standard, and a study can score five and still be false, so the phone call to the named customer is the real test. When two agencies tie, prefer the one whose study admits a limit.
Weak and strong, side by side
Here are two invented extracts, made up to show the difference. Neither is a real client.
| Weak | Strong | |
|---|---|---|
| Starting point | "A struggling retailer" | "A four-person fastener shop in Rajkot, enquiries by phone only, none written down" |
| What changed | "Optimised their digital presence" | "Listing corrected, quotes itemised, one person made responsible for callbacks" |
| Result | "Sales shot up massively" | "Twelve quotes chased in the first stretch, seven turned into orders, counted from the shop's own sheet" |
| Verified by | Not stated | "The owner's sheet, shared with us" |
| Honest limit | None | "Walk-in sales were not tracked, so the full effect is unknown" |
The strong column is smaller and less exciting, and far more believable.
Warning signs that need no calculation
- A percentage with no base number.
- A time period that is missing ("in just a few months").
- Logos of big companies with no sentence about the work done for them.
- Quotations from customers with first names only and no business.
- Every case ending in a dramatic jump. Real results are often slow and uneven.
- Stock photographs where the customer's premises should be.
For example: a factory owner in Rajkot choosing an agency
Take an invented owner of a bearing-housing unit in Rajkot, shortlisting two agencies. Agency A shows three case studies, each with a large percentage and a logo. Agency B shows one, about a fabrication unit in a nearby town.
The owner scores them with the six questions. Agency A's studies each answered one: they named a customer. None gave a starting figure or said how the result was counted. Agency B's study answered five. It described what the fabricator had before, named the three changes, said the figures came from the fabricator's own order sheet, and admitted that a directory listing had produced nothing.
The first instinct is to pick A, because three look better than one. The scoring changes that. The owner rings the fabricator named in B's study and asks one question: "What did they do that you would not have noticed?" The answer is specific, and it settles the choice, though B quotes higher and cannot start for a month. Three impressive logos told the owner less than one honest account and one phone call.
When a case study is still not enough
A good case study shows that something worked for someone. It does not show it will work for you. The customer may have had a better product, a lucky season or a bigger budget. So use it to decide who goes on your shortlist, then test with a smaller first piece of work and a baseline you agree beforehand.
Be careful with industries where rules limit what may be claimed, such as health, law or finance. A case study there may be restricted by the professional body, and a missing number can be a sign of compliance, not concealment. Ask, do not assume.
Score one case study tonight
Take the study that most impressed you, from any supplier, and score it on the six questions. Then ring the named customer if there is one. The same questions work on the studies in our case studies, and if one leaves a question unanswered, ask us. The work of writing them for clients is the customer case studies service. If you are comparing suppliers and want a second opinion on what they showed you, tell us what you need.
Related services
