Ask for the testimonial two to four weeks after delivery and the case study at six to twelve weeks, because a testimonial needs the experience to be fresh and a case study needs results to exist. Always send a draft quote for them to edit rather than a blank request - it roughly triples the response rate and it is not putting words in their mouth if you are quoting what they actually said. A case study follows Situation, Trigger, Barrier, Solution, Results, and it leads with the number in the headline. The reason most agencies have thin proof is not that clients refuse; it is that nobody asks at the moment the client is most willing.
Most agencies have a portfolio and almost no proof.
The portfolio shows what the work looked like. It does not answer the question a prospective buyer is actually asking, which is not "can you design?" but "does hiring you work out?" That question is answered by other people's experience, and the only way to have that on your site is to have collected it.
The collection almost never happens, and the reason is timing rather than reluctance. Agencies ask nine months later, when a case study is needed for a pitch, by which point the champion has moved on, the numbers are buried, and the enthusiasm has cooled into polite vagueness.
This guide covers when to ask, exactly how to ask so people say yes, the difference between a testimonial and a case study, the structure that makes a case study persuasive, how to get results data out of a client who is not tracking it, and what to do when they say no.
Why the proof is thin
Three reasons, and none is that clients dislike you.
Nobody owns it. Testimonials sit between delivery and marketing. The delivery team has moved on; the marketing person was not in the relationship. So it becomes something that happens when someone remembers, which is never.
The ask is too big. "Would you write us a testimonial?" is a request for someone to draft, edit and approve a piece of writing about a supplier. It is a task, it goes on a list, and it stays there.
It is asked at the wrong time. Either on delivery day, before the client can speak to outcomes, or nine months later, when there is nothing vivid left to say.
Fix those three and the collection rate goes from roughly one in five engagements to most of them.
Testimonial versus case study
Two different assets, different jobs, different timing. Conflating them is why agencies end up with neither.
A testimonial is one to three sentences of opinion, attributed. It answers "what was it like working with them?" It costs the client five minutes. It can be collected from almost every engagement.
A case study is a structured story with numbers. It answers "does this work, for a company like mine?" It costs the client an hour or more of involvement, needs their approval, and only works when there are results worth reporting.
Collect testimonials from everyone. Build case studies from the small number of engagements where the outcome was measurable and good.
When to ask
The timing is specific and it is most of the game.
Testimonial: two to four weeks after delivery. Late enough that they have lived with the work and can say something about the outcome. Early enough that the experience of working with you is still concrete - they remember the specific moment you caught something, or turned something round quickly.
Case study: six to twelve weeks. Results need time to exist. A case study built at four weeks says "the site launched"; the same one at ten weeks says "demo requests are up 34%."
Referral: after both. A client who has just given you a quote and reviewed a case study has, in their own words, made the argument for you twice. Asking then is a much smaller step than asking cold.
Signal the testimonial ask during the wrap-up call, so the email is expected rather than cold. Our guide to client offboarding covers where these fit in the wider ending process, which is the thing that makes them systematic rather than occasional.
How to ask for a testimonial
The single highest-leverage change: send a draft.
Hi Sarah,
Now the site has been live a few weeks and the numbers have settled, would you be happy for us to quote you on our site?
To save you writing anything, here is a draft based on what you said on our wrap-up call - please edit it however you like, or bin it and write your own if you would rather:
"We came to them with a site our sales team apologised for. Ten weeks later demo requests were up by a third. What stood out was that they pushed back on ideas that would not work rather than just building what we asked for."
Happy for it to be attributed to you and Northwind, or anonymised to "a B2B software company" - whichever suits.
Four things are doing work.
It arrives at the right moment. Covered above.
It removes the blank page. This is the mechanism. Most people will edit a draft in two minutes and send it back the same day; the same people will not write 60 words from scratch for six weeks.
It quotes what they said. Not invention - you are drafting from the wrap-up call. That is why taking a note of the good sentences during that call matters: your best testimonials are usually things clients said in passing.
It offers the anonymised option. A meaningful number of people work somewhere with a policy about endorsing suppliers. Without an alternative they simply do not reply, and you never learn why. With one, you get the quote.
Send it from the person they worked with, not a marketing address. That alone changes the response rate.
What makes a testimonial useful
Most testimonials are useless because they are generic. "Great to work with, highly recommend" tells a prospect nothing they did not assume.
Useful ones do at least one of these:
Name the objection they had. "We were worried about handing this to an outside team" is worth ten adjectives, because the reader has the same worry.
Contain a number. Any number. Even "we cut our reporting time from two days to two hours."
Describe a specific behaviour. "They told us when an idea of ours was wrong" is credible precisely because it is not flattering in a generic way.
Come from someone identifiable. Name, role, company. Advice on B2B case studies from The Simons Group makes the point that attribution is what makes proof credible - an unattributed quote reads as invented, fairly or not.
When you draft, aim at one of those four. If the client edits it into something generic, you can gently offer the original back: "Would you be comfortable keeping the line about the sales team? It is the bit that will mean most to someone in the same position."
The case study structure that works
The reliable shape, and it is not chronological:
1. Headline with the result. Lead with the number. "How Northwind increased demo requests 34% in one quarter" rather than "Northwind website redesign." A reader deciding whether to spend three minutes needs the outcome first.
2. A snapshot box. Industry, company size, what they bought, headline metrics. Above the fold, scannable. Most readers will read only this, and that is fine - it is doing its job.
3. Situation. Where they were. Brief and factual.
4. Trigger. What made this urgent now. This is the section most case studies skip and the one that makes a reader recognise themselves - people do not act on chronic problems, they act on triggers. A funding round, a competitor, a board commitment, a failed launch.
5. Barrier. What made it hard, or what they tried that did not work. This is what stops it reading as an advert. A story with no obstacle is not a story.
6. Solution. What you did, in plain language. Shorter than you want it to be. The reader cares less about your process than you do.
7. Results. Numbers, with a baseline and a timeframe. "34% increase" means nothing without "from 1.2% to 1.6% conversion, over the quarter following launch."
8. Quote. In their words, ideally addressing the barrier from section five.
Omniscient Digital's guide to high-converting B2B case studies and Media Shower's best-practice piece both converge on broadly this problem-solution-outcome spine, and Webstacks' collection of examples is useful for seeing how the snapshot box works in practice.
Keep the whole thing to 600-900 words. Case studies are skimmed, not read, and a long one simply gets skimmed harder.
Getting the results data
The hardest practical part, because clients frequently are not tracking the thing you improved.
Establish the baseline before you start. This is the fix, and it has to happen at kickoff rather than at case-study time. Ten minutes at the start recording current conversion, current time-to-publish, current whatever - is the difference between a case study with numbers and one without. Agencies that do this consistently have dramatically better proof, and the reason is not better writing.
Ask for the number as a favour, not a requirement. "Would you mind pulling the demo request numbers for the last quarter? It would let us write this up properly - and it is useful for you either way."
Offer to pull it yourself if you still have analytics access. Frequently faster for everyone.
Use their proxy if they have no metric. Time saved, headcount avoided, a process that used to take two days. Softer than revenue and much better than nothing.
Use a qualitative result if there is genuinely no number. "The sales team now sends the site to prospects before calls, which they never did before" is a real outcome. Label it as what it is rather than dressing it up.
Never invent or round generously. A fabricated number is the one mistake that cannot be recovered from, because the client will read the case study, and so will people who know them.
When they say no
Some will, and the reason usually determines the workaround.
Company policy on endorsements. Common in finance, legal, healthcare and the public sector. Offer anonymisation: "a mid-sized insurance broker" plus the numbers is still useful proof. Many people who cannot be named can approve an anonymous version.
Competitive sensitivity. They do not want competitors knowing what they are doing. Offer to delay by six or twelve months, or to omit the specific tactics and keep the outcome.
They are simply too busy. Almost always the real answer. This is what the pre-drafted quote solves; if they still do not reply after one follow-up, let it go and ask again at the six-month check-in.
The engagement did not go well enough. Do not push. Ask for feedback instead, sincerely, and take it seriously - that conversation is worth more to the business than the testimonial would have been.
One follow-up, then stop. Chasing a testimonial damages the relationship that produced it, and the relationship is worth more.
Building the habit
Four things to set up once.
A note-taking habit on wrap-up calls. Write down the sentences clients say about working with you, verbatim, as they say them. This is where drafted quotes come from, and it takes no additional time.
A baseline capture at kickoff. One field, filled in at the start of every engagement, recording the metric you expect to move and its current value.
Two email templates. The testimonial request with a placeholder for the drafted quote, and the case study request framed as joint content.
A scheduled trigger. The T+2 week and T+8 week touchpoints need to exist as tasks with an owner and a date, created when a project moves to its final phase. Every step here is easy; all of them fall outside the window when anyone is naturally paying attention, which is exactly why they need to be scheduled rather than intended.
Video, written, and which is worth the effort
A question worth answering deliberately, because video testimonials consume far more of everyone's goodwill than written ones.
Written quotes cost the client two minutes, can be collected from nearly every engagement, and are easy to place anywhere. Start here and get comprehensive coverage before considering anything else.
Video is more persuasive per unit and dramatically more expensive to obtain. It needs scheduling, a client willing to be on camera, decent audio, and editing. Realistically you will get one for every fifteen written ones, and only from your warmest relationships.
The pragmatic middle: record the wrap-up call (with permission) and ask one or two testimonial-shaped questions at the end. "If someone in your position asked whether this was worth doing, what would you tell them?" You get a usable 40-second clip from a call that was happening anyway, with no additional imposition on the client. That single question, asked at the end of every wrap-up, produces more video proof than any dedicated campaign.
Screenshots of unsolicited praise are underrated. The message a client sent at 9pm saying the launch went well is more credible than any polished quote, precisely because it was not solicited. Ask permission before using it, always.
The proof you already have and have not collected
Before running any new process, most agencies have unclaimed proof sitting in their inbox.
Search your email and chat history for phrases like "thank you", "brilliant", "exactly what we needed". You will find a dozen genuine, specific compliments from the last two years. Every one of those is a testimonial waiting for a permission request.
The email to send is short:
This is a bit of a blast from the past - you sent this after we launched the portal last year and it stuck with me. Would you be comfortable with us using it as a quote on our site? Happy to trim it or attribute it however you like.
Response rates on these are high, because you are not asking them to produce anything. They already wrote it.
Check your project history for outcomes you never followed up. Projects that shipped and then vanished from your attention frequently produced results nobody asked about. A short, genuinely curious message - "how did that end up performing?" - sometimes yields a number good enough to build a case study around eighteen months later.
Keeping proof current
Testimonials age, and old proof is worse than it looks.
A quote attributed to someone who left that company three years ago, referencing a service you no longer emphasise, quietly signals that your best work is behind you. Two habits keep the set fresh:
Date them, at least internally. Know how old each piece of proof is, even if the date is not published.
Retire and replace annually. Once a year, review what is on the site. Anything older than about three years should be earning its place - either it is exceptionally strong, or it goes. A page of eight recent, specific quotes beats a page of thirty spanning a decade.
This also gives the collection process a purpose beyond accumulation: you are maintaining a set, not filling a page.
Getting approval without friction
The step that stalls most case studies is not writing them - it is getting sign-off from a client organisation that has a process nobody mentioned.
Ask about approval at the same time as you ask for the case study. "Is there anyone else who would need to see this before it goes out?" Discovering the legal review at draft three costs weeks.
Send a finished draft, not a request to participate. A client asked to contribute to a case study has acquired a project. A client sent a complete draft with a request for corrections has a ten-minute task.
Offer three levels of attribution in the same message: fully named, company named but person anonymous, or fully anonymised. A meaningful number of people who cannot get named approval can approve one of the other two, and without the option they simply do not reply.
Give a soft deadline. "If I don't hear back by the 20th I'll assume the timing isn't right and park it" removes the obligation, which reliably increases the response rate rather than reducing it.
The one habit that matters most
If you adopt nothing else from this guide, adopt this: write down the sentences clients say about working with you, verbatim, as they say them.
On calls, in messages, in passing at the end of a review. A running note, thirty seconds each time.
That single habit solves the hardest part of every testimonial request, because the draft you send back is built from their own words rather than your invention - which is what makes people comfortable approving it, and what makes the resulting quote specific rather than generic.
Where the proof actually gets used
Worth being deliberate about, because collected proof frequently sits unused on a page nobody visits.
In the proposal. One relevant testimonial or a two-line case study summary, placed near the pricing. This is the highest-value placement on the list and the one most often missed - a buyer reading your price is the buyer most in need of reassurance. Our guide to writing a proposal covers where it fits.
On the specific service page, not just a testimonials page. Proof next to the thing it is proof of.
In the first sales conversation, spoken. "We did something similar for a company in your position - they had the same worry about handing it over."
In the follow-up after silence. A short case study is a legitimate reason to make contact that is not "just checking in."
A testimonials page collecting everything is fine as an archive. It is not where proof does its work.
The summary
The agencies with strong proof are not luckier or better liked. They ask at two and eight weeks instead of nine months, they send a draft instead of a blank request, and they wrote down the baseline metric on day one so there is a number to report.
All three are small. The compounding effect is that after two years you have twenty pieces of specific, attributed, numerical proof, and your competitors have a portfolio.