Agency Case Study Teardown: 6 Examples, Annotated

Agency case studies prove judgement, not a tool. These six show how to handle the time window, the before, the unnamed client and whose voice carries the result.

Agency Case Study Teardown: 6 Examples, Annotated
Image URL
Do not index
Do not index
AI summary
Agency case studies differ from product ones by emphasizing time windows, before‑after comparisons, and the client’s voice; they should include clear numbers with baselines, attribution words, visual or textual “before” elements, decision‑focused narratives, and properly attributed client quotes, while avoiding agency‑only claims and overly generic service lists.
Title
Agency Case Study Teardown: 6 Examples, Annotated
Date
Oct 11, 2026
Description
Six live agency case studies from Single Grain, Power Digital, Directive, Ramotion, WebFX and Pentagram, annotated: the time window, the before, the client's voice.
Status
In progress
Current Column
In the queue
Person
Writer
Single Grain's SmartRent case study measures its result over one month. Power Digital's Casper case study measures its result over two years. Both are agency case studies, both are live, and the gap between those two time windows is most of what makes agency proof different from the product kind.
A SaaS case study has to prove a tool works. An agency case study has to prove a team's judgement worked, for a client the reader has never met, over a stretch of time the reader is about to pay for. So it carries things a product story can skip: a timeline, a "before," an argument for why the agency's thinking made the difference, and a client who has to agree to all of it in public.
This teardown looks at six live agency case studies, all checked on 11 October 2026, and annotates how each one handles those jobs. If you want the full picture of how case studies get made, start with our guide to customer case studies. For where they sit next to quotes, videos and reviews, see the whole testimonial pipeline.

Five things to check in any agency case study

  1. Whose voice carries the result. The client's, or the agency's own staff?
  1. Whether the client is named. And if not, how specifically they're described.
  1. The number. Does it come with a time window, a baseline, and an honest word about how much of it the agency can claim?
  1. The "before." Shown as a picture, listed in text, or skipped?
  1. The expertise. Argued through decisions the agency made, or just asserted?

1. Single Grain × SmartRent: the fast result, with its baseline

Single Grain's SmartRent case study has the result in the headline: "How SmartRent Achieved 10 Monthly SQLs From LLMs." The first paragraph adds the time window, "136 new AI citations within 30 days," and the results block puts a dollar figure on it: "10 SQLs per month," "$126,000," "worth approximately."
The best number on the page is the smallest one. Under "Google AIO" it says "50%," then "From 64 to 96 in total citations." That's a percentage with its baseline attached, so a reader can check the maths in their head. The 136 and the $126,000 don't have that.
The method is a numbered framework: a "5-tier priority framework" carried out as seven named actions, from "Technical Website Audit" to "Backlinking & Web Citation Strategy."
What it costs: a one-month window is persuasive and fragile. A reader who has bought SEO before will ask what month two looked like. And there's no SmartRent voice on the page. Every claim comes from Single Grain.
Copy: the "from 64 to 96" pattern on every percentage. Skip: a value estimate like "worth approximately $126,000" unless you can say how it was worked out.
Single Grain's SmartRent case study, annotated: the baseline, the "worth approximately" qualifier and the 30-day window. Screenshot: Single Grain, 11 Oct 2026.
Single Grain's SmartRent case study, annotated: the baseline, the "worth approximately" qualifier and the 30-day window. Screenshot: Single Grain, 11 Oct 2026.

2. Power Digital × Casper, and the client it can't name

Power Digital's Casper case study sits at the other end of the clock. The H1 reads "A Succinct SEO Strategy Drives Global Expansion and $100M+ In Sales Over 2 Years," the page is dated October 30, 2024, and the two result tiles are "188% Increase in organic traffic year-over-year" and "$100M+ Sales over 2 years, supported by strong SEO."
Look at the word "supported." An SEO agency didn't sell $100M of mattresses, and the tile doesn't pretend it did. The body says the same thing in longer form: the traffic growth "contributed significantly to" the sales. That's the attribution word doing honest work, and it makes the big number easier to believe.
The page also tells the story in stages a buyer can map onto their own engagement: a technical audit first, then keyword targeting, then becoming "Casper's primary point of contact for all SEO vendors," then international launch in Canada, Germany, Austria, France, Switzerland and the UK.
Now open Power Digital's "Bold Underwear Brand" case study. Same agency, same template, but the client isn't named anywhere. Results are "+27% YoY Revenue" and "+13% YoY Average Order Value."
What rescues it is the description. The brand was founded in 2007, rose after "a notable media moment," and is "a dominant player in the dual-gender underwear market." You can't name it, but you can picture it, and a buyer from a similar brand will recognise themselves.
What it costs: the anonymous version has no quote and no face, so the numbers carry everything. And its services row runs from Affiliate to TikTok, which makes it hard to tell which one moved the 27%.
Copy: "supported by" or "contributed to" when the result has more than one cause, and a specific description when the name is off-limits. Skip: the full services list on an anonymous story. Name the two that mattered.
Power Digital's Casper case study and its Bold Underwear Brand page, annotated. Screenshot: Power Digital, 11 Oct 2026.
Power Digital's Casper case study and its Bold Underwear Brand page, annotated. Screenshot: Power Digital, 11 Oct 2026.

3. Directive × Seagate: the agency tells the story

Directive's Seagate case study opens with three numbers: "75% Increase in Organic Traffic to Blog Subfolder MoM," "50.6% Improvement in the Average Position of Core Keywords," and "45.2% Increase in Organic Traffic to Target Segments."
The pull quote below them is about as enthusiastic as these get: traffic climbed to "more than triple what we had been seeing." It's signed "David Waiter, SEO Strategist @Directive." The person vouching for the result works at the agency.
The body is where the page earns its keep. It's honest about the slow part. "Results take time and consistency," it says, and it describes going "through multiple iterations of the content workflow" before anything moved. It gives one concrete output, "60 pieces in Q2 of 2022 alone," and one concrete division of labour: Directive ran Seagate's Wrike workspace, "leaving only approvals for the client."
That last line is the expertise claim done well. It doesn't say "we're experts." It shows a decision about who does what, which a buyer can picture in their own team.
What it costs: Seagate never speaks. And "MoM" on the first tile is a monthly comparison with no month attached, so a reader can't tell when it happened or whether it held.
Copy: the paragraph that admits how long it took, and the line about who did what. Skip: a staff quote in the slot where a client quote belongs.
Directive's Seagate case study, annotated: the MoM label with no month and the agency byline on the quote. Screenshot: Directive, 11 Oct 2026.
Directive's Seagate case study, annotated: the MoM label with no month and the agency byline on the quote. Screenshot: Directive, 11 Oct 2026.

4. Ramotion × Clearbit: before and after, as pictures

Design agencies have an advantage the others don't: the deliverable is visible. Ramotion's Clearbit case study uses it. Alongside six embedded videos, the page carries an image labelled "Old Clearbit website" in its alt text. The before is shown, not described.
The fact row up top is a different kind of number. It reads "Exit $150M" over "Acquisition by Hubspot," then "B2B Data 50M" over "Company records," then "Usage 1500" over "Business customers." Those describe how big and successful Clearbit is. None of them measures the redesign. Next to them sit a "Services" line that reads "Web design" and a "Tech stack" line, "Next.js, Tailwind CSS, Chart.js, Contentful," which is a detail written for whoever has to maintain the site afterwards.
The outcome section is in words, not figures: "improved conversion rates, lower bounce rates, and enhanced page speed." The client quote is where the proof sits. It's from Nick Wentz, VP Marketing at Clearbit: "We partnered with Ramotion to redesign our entire marketing site from the ground up—new pages, messaging, graphics, and visual identity," and later, "Operating under an extremely tight timeline," the team "delivered at every phase of the project."
What it costs: the borrowed numbers. A reader skimming the fact row can easily take "$150M" as something the website did. The page itself links the two, saying the work "propelled Clearbit to a successful merger with HubSpot." If you put client-scale numbers next to your work, label them clearly as client facts.
Copy: the before picture, and a named client with a job title. Skip: stature numbers styled like results.
Ramotion's Clearbit case study, annotated: client-scale facts, the before image and the named, titled quote. Screenshot: Ramotion, 11 Oct 2026.
Ramotion's Clearbit case study, annotated: client-scale facts, the before image and the named, titled quote. Screenshot: Ramotion, 11 Oct 2026.

5. WebFX × Asbury Electric: the before, as a list

Not every agency deliverable is something you can screenshot. WebFX's Asbury Electric case study is local SEO and paid search for an electrician in Williamsburg, Virginia, so the before/after is done in text.
Two columns: "Before WebFX" lists four problems, from "Underperforming search engine optimization (SEO)" to "Poor digital marketing results reporting and decision-making." "After WebFX" answers with three changes. It's the plainest version of the pattern, and it works because the reader can match their own list against it.
The page also says how the client found them: "After reading our online reviews and enjoying their initial consultation." That one sentence is useful to a buyer who's at the same step right now.
At the bottom it names the team, three people with their departments, and a line that reads "502+ hours dedicated to driving results." Hours is an unusual thing to publish, and for a small business wondering whether it'll get real attention, it's a reasonable answer.
What it costs: the headline stat, a "50%" over the label "YOY INCREASE IN IMPRESSIONS FROM LOCAL SEO," is an early-funnel number. Impressions aren't calls or jobs. The client quote is signed "Asbury Electric" with no person attached, and the same quote appears twice.
Copy: the two-column before/after when your work isn't visual, and the "how they found us" line. Skip: a company-only signature. Ask for a name and a role, and if the client prefers to stay anonymous, say so on the page.
WebFX's Asbury Electric case study, annotated: the before/after columns, the impressions metric and the company-only signature. Screenshot: WebFX, 11 Oct 2026.
WebFX's Asbury Electric case study, annotated: the before/after columns, the impressions metric and the company-only signature. Screenshot: WebFX, 11 Oct 2026.

6. Pentagram × YouMe Kyoto Triennale: no numbers at all

Pentagram's YouMe Kyoto Triennale page is a case study for an event that hasn't happened yet. The page says the Triennale "will open its doors for the first time in October 2026." There's no result to report, so there's no number.
What it has instead is the reasoning, step by step. A manifesto came first. Then Pentagram received "a collection of hand drawn cards from local Kyoto artists and craftspeople." Then, in 2025, Partner Harry Pearce presented the idea to the Triennale committee: a symbol based on "a traditional interlocking three-part Japanese timber joint," built to be shown "empty as well as solid, filled with images, textures or lines." The people are named, down to Associate Partner Johannes Grimmond.
This is the "we're the expert" framing with no adjectives. The page never says Pentagram is good at identity work. It walks you through the decisions and lets the work make the case.
What it costs: it only works when the buyer can judge the work by looking at it, and when the agency's name already does some of the selling. A performance agency publishing a case study with no numbers would be read as hiding them.
Copy: the sequence of decisions, with names on them. Skip: the no-numbers approach if your client is buying an outcome rather than an artefact.
Pentagram's YouMe Kyoto Triennale page, annotated: no metrics in the header, the named partner credit and the process paragraph. Screenshot: Pentagram, 11 Oct 2026.
Pentagram's YouMe Kyoto Triennale page, annotated: no metrics in the header, the named partner credit and the process paragraph. Screenshot: Pentagram, 11 Oct 2026.

What agencies do that product companies don't

Put these six next to the SaaS pages in annotated case study landing pages and four differences stand out.
  • The time window is part of the claim. One month (Single Grain), two years (Power Digital), a whole quarter's output (Directive), a launch date that hasn't arrived (Pentagram). A buyer hiring an agency is buying time, so the case study has to say how much.
  • The "before" is structural. Ramotion shows it as an image, WebFX as a list, Directive as a paragraph about the workflow that didn't work. Product case studies often skip it.
  • The agency is a character in the story. Named strategists, named partners, a named account team. That's a strength until the agency's voice replaces the client's, which is what happens on the Directive and Single Grain pages.
  • Every CTA is "hire us." "Explore Partnership" and "Your success story is waiting for you." on Directive. "Work With Us" on Single Grain. "Contact WebFX today" on WebFX. "Let's talk" on Ramotion. There's no free trial to offer, so the case study carries the whole sales job.

Writing your agency's first one

Here's what I'd do, in this order.
  1. Get the client's words before you write yours. The weakest moment on most of these six pages is the client's voice: a quote from the agency's own strategist, a quote from a company name with no person, or no client quote at all. Ask the client what changed, in their own words, and put that above anything your team says.
  1. Give every number a window and a baseline. "From 64 to 96" beats "50%." "Over 2 years" beats "year-over-year." If you only have the percentage, keep looking.
  1. Pick an honest attribution word. "Supported by," "contributed to," "alongside." Buyers know you weren't the only cause. Saying so makes the rest of the page easier to trust.
  1. Show the before. A screenshot if your work is visual, a two-column list if it isn't.
  1. Write the decisions, not the service list. "Leaving only approvals for the client" tells a buyer more about working with you than twelve service tags.
  1. Get sign-off on the name, the logo and the numbers separately. Clients who'll happily give a quote often won't publish revenue. If they want to stay anonymous, describe them as specifically as Power Digital does, and say on the page that the client isn't named at their request. The legal side of testimonials covers what to get in writing.
If the story itself won't come together, case study formats covers the structure, and testimonials for agencies covers the shorter proof that sits next to it.

Doing this with Testimonial

The hardest part of every page above is the part the agency can't write itself: the client's own account of what changed. Agencies usually get it on a call, then paraphrase it, which is how a strategist ends up signing the quote.
Case Study sends your client a set of 5 to 10 questions they answer in their own time, by typing or by voice, with no call to schedule and no signup. Voice answers are transcribed, the original audio is kept, and responses arrive in your inbox already attributed to the person you invited. Ask "What was the situation before we started?", "What number changed, and over how long?" and "Can we use your name and title?" and you have the baseline, the time window and the permission in one go. Running a case study interview without a call walks through the full question set.
Case Study is on the Ultimate + plan, priced per space. Details are on pricing.

Wrapping up

Six agency case studies, six different answers to the same problem: proving judgement to someone who wasn't there. The habits worth stealing are a time window on every number, a before the reader can see, an attribution word that doesn't overclaim, decisions instead of service lists, and the client, not your team, saying what changed.
Writing one this week and not sure which number to lead with? Send me what you've got and I'll tell you which one I'd put first.

Written by

Damon Chen
Damon Chen

Founder of Testimonial