How-to8 min read

How to turn customer evidence into case studies

The evidence already exists. This is how to turn it into proof you can use.

To turn customer evidence into a case study, start with the source you already have: a public review, an existing case study, a website page or a customer call transcript. Pull out the passages where the customer describes the situation before, what changed and any figure attached; write from those passages rather than filling gaps; then get the exact wording confirmed and keep a record of what was approved. The two steps teams skip are source tracing and the review date — without the first you cannot defend the claim, and without the second the case study quietly stops being true.

1. Start with the source you already have

Start by gathering the places where customers already describe the value they received: reviews on Google, Trustpilot, G2 or Clutch, published case studies, website testimonials and customer call transcripts. If the source is a call, you do not need a dedicated interview; useful proof often appears in ordinary renewals, check-ins and quarterly reviews when someone asks questions like these:

  • "What was this like before?" — you cannot show a change without the starting point.
  • "How would you describe the difference to someone in your role?" — this is where quotable lines come from.
  • "Is there a number attached to that?" — hours, percentages, headcount, cycle time. Ask once and accept no.
  • "Who else internally felt the difference?" — it tells you whose approval you'll need later.

For calls, make sure the conversation is transcribed. Speaker labels make accurate attribution possible. For a public review, keep the reviewer name, platform and link to the original. For existing content, keep the page URL or source file. In every route, preserve enough context to trace the claim back to where it came from.

2. Extract the proof, not the whole source

Go through the review, case study, website page or transcript and pull out only the passages that carry proof. In practice that is four kinds of line: the before state, the change, a figure and a direct quote worth using verbatim. Everything else is context, not evidence, and should not be used to strengthen a claim the source does not support.

This is also the point to be strict about attribution. If you can't tell who said a line, don't use it as a customer quote.

3. Write it in the customer's words

A case study a buyer believes is short, specific and clearly the customer's account rather than yours. A structure that consistently works:

  1. Who they are, in one line, specific enough that a similar buyer recognises themselves.
  2. The situation before, in concrete terms — what was slow, manual, unreliable or expensive.
  3. What changed, and the figure if there is one.
  4. One or two direct quotes, verbatim, not tidied into marketing voice.
  5. What they're doing next, which signals the relationship is live rather than historical.

Resist adding claims the source doesn't support. If the customer said "a lot faster", write "a lot faster" — an invented percentage is the fastest way to lose trust and an approval.

4. Get approval you can actually prove

Two separate gates matter here. Internal approval means your team is happy with the draft. Customer approval means the named customer has read the exact wording and agreed to it being published. Verbal agreement on a call is not the second one.

  • Send the specific wording, not a description of it.
  • Ask explicitly what they're approving: the quote, their name, their company name, the logo, the figure.
  • Record who approved, when, and which version they saw.
  • If you edit after sign-off, treat the approval as no longer covering the change.

That last point is where most published proof drifts out of compliance: a small edit months later, nobody re-checks, and the wording on your site is no longer the wording anybody agreed to.

5. Give every case study a review date

Proof perishes. Figures move, the champion leaves, the customer switches plans, and occasionally they stop being a customer altogether. A case study from two years ago with a hard number in it is a liability unless somebody has checked it since.

What to check, and roughly how often.
TriggerWhat to checkAction
12 months since approvalIs the figure still accurate?Ask the customer to confirm or correct
Champion changes roleIs the quote still attributable?Re-confirm with their successor
Customer churnsCan you still use it?Withdraw or mark as historical
Wording editedDoes approval still cover it?Re-approve the new version

One link asking the customer to confirm, correct or withdraw is far easier to send than a fresh interview request, and it keeps the whole library defensible.

Doing this with Casily

Casily automates the mechanical parts. It brings in reviews from Google, Trustpilot, G2 and Clutch, existing case studies and website content, and transcripts from customer calls. It extracts the proof-worthy passages, keeps figures and quotes tied to their sources, and drafts material for review. Customer confirmation is sent as a link to the exact wording and bound to that version, so a later edit is flagged rather than carried forward. Ageing proof and moved figures are surfaced as work, with one link to ask the customer to confirm, correct or withdraw.

Raw transcripts don't have to stick around: proof-relevant passages are extracted, then the transcript is discarded, kept seven days, or retained, per workspace setting, defaulting to seven days.

If you'd rather try one step by hand first, the mini case study generator and the pull-quote extractor are free and need no signup.

Frequently asked questions

Can I write a case study from a call the customer didn't know was for marketing?

You can draft from it, but you cannot publish it without going back and getting explicit approval of the wording, the name and any figure. Approval of the published text is the gate, not consent to the call itself.

What if the customer won't give a number?

Publish the change without inventing a figure. A specific, believable description of what improved is stronger than a number the customer never agreed to and might dispute later.

How long should a case study be?

Short enough that a buyer reads all of it — typically 300 to 600 words with one or two verbatim quotes. Length is not evidence; specificity is.

How often should we re-check published proof?

At least once a year for anything with a figure in it, and immediately when the customer's champion changes role or the account status changes. Casily flags ageing items so the check doesn't rely on someone remembering.

Do we need to keep the transcript to keep the case study?

No. Once the proof-relevant passages are extracted, the raw transcript can be discarded, kept seven days, or retained, per workspace setting. The proof and the passages it came from stay either way.

Keep reading