A credible marketing case study is built from project records before it is shaped into a story. The public article may feel simple, but every statement about scope, change, performance, or client experience should trace back to a source that existed before publication.
That discipline solves two common problems at once. It keeps the narrative from claiming more than the work can prove, and it gives sales, marketing, and account teams a durable record they can revisit when a quote ages, a metric definition changes, or the client asks where a number came from.
The practical method is to maintain one evidence card per public claim. The card records the wording, claim type, source, time window, calculation, limits, approver, and review date. The writer can then tell a clear story without asking a screenshot, dashboard, or testimonial to prove something it cannot.
The evidence file comes before the narrative
A project folder usually contains plenty of material: proposals, briefs, designs, launch notes, analytics exports, emails, meeting recordings, and client feedback. A case study does not need all of it. It needs the subset that supports the exact public statements the business wants to make.
The FTC's Policy Statement Regarding Advertising Substantiation says advertisers and agencies should have a reasonable basis for objective express and implied claims before they are disseminated. For editorial operations, that means the evidence review belongs upstream of the headline, not in a frantic approval pass after the story is written.
Start by writing the proposed claim in one sentence. Then ask what a reasonable reader would understand it to mean. “We rebuilt the website” is a delivery fact. “The new website increased qualified leads” is a performance and causation claim. The second sentence needs much more than proof that a launch happened.
The Branding Bull's case-study portfolio spans brand, web, growth, social, and product work. A multi-service story can be useful without pretending one deliverable deserves all the credit. The evidence file preserves which work occurred, which results were observed, and which connection is actually supported.
Give every public sentence a claim type
Claim types prevent a smooth paragraph from collapsing unlike evidence into one sweeping result. Use a small vocabulary that reviewers can apply consistently:
- Project fact. A dated, documented statement about the scope, team, launch, deliverable, or operating change.
- Observed change. A measured difference between defined periods or groups, without a claim that the project alone caused it.
- Attributed result. A conversion, visit, lead, or revenue value assigned by a named platform and attribution model.
- Incremental result. A causal estimate supported by an appropriate experiment or comparison design, including its limits.
- Endorsement. A client's approved opinion or experience, kept distinct from the agency's own factual and performance claims.
The label is not public copy. It is a control inside the evidence file. Once the type is visible, the reviewer can demand the right source instead of approving a sentence because it sounds plausible.
An observed result is not automatically a caused result
Marketing platforms answer different questions. An analytics or advertising report can show that a model assigned credit to a channel or interaction. That is valuable evidence, but it does not by itself show what would have happened without the work.
Google's current explanation of data-driven attribution describes how conversion credit is distributed using advertiser-specific path data. Its separate Conversion Lift measurement guidance distinguishes standard attributed conversions from incremental conversions estimated with treatment and control groups. Record which one the case study actually has.
A concise claim card should include the metric definition, source property or account, date range, comparison window, filters, attribution model and window, calculation, simultaneous changes, known data gaps, and the file or query that reproduces the number. If a percentage is published, store the underlying counts too.
This distinction is visible in the published work itself. The Redtail Telematics case study covers social content, video, web, paid search, product experience, and event support. A result reported across that combined work should remain framed as combined-work evidence unless a narrower source supports a narrower claim. The MYNET Insurance case study demonstrates the other useful option: explain the documented foundation and operating change without inventing a performance number.
The strongest wording is often the narrowest accurate wording. “Qualified inquiry rate rose from the documented baseline during the launch period” can be both more credible and more useful than “the redesign transformed lead generation,” especially when paid spend, offer, routing, or seasonality changed at the same time.
Keep client words on a separate evidence track
A client quote is not a substitute for the project record. It is evidence of that client's stated experience. Preserve the recording, transcript, email, or survey response; the edited draft; the context in which it will appear; the name and role the client approved; and the date of approval.
The current FTC Endorsement Guides in 16 CFR Part 255 say endorsements must reflect honest opinions or experiences and cannot convey a representation that would be deceptive if the advertiser made it directly. They allow editing, but not rewording that distorts the endorser's opinion or experience.
Do not let quotation marks turn a polished paraphrase into supposed verbatim speech. If the wording has been condensed, get approval for the final public quote and retain the source. If the client speaks only about responsiveness, do not use the quote to imply a sales result.
The FTC's 2023 Endorsement Guides Q&A also warns that exceptional-result testimonials can imply what others should expect, and that generic “results may vary” language does not fix that impression. Its Consumer Reviews and Testimonials Rule Q&A is narrower in scope, so route legal questions to qualified counsel rather than assuming every B2B case study fits the same rule. The editorial rule is simpler: never fabricate a person, experience, quote, or approval.
Turn the file into a readable story without hiding the limits
Evidence does not require a sterile case study. Open with the client's real operating constraint, show the decisions that changed the work, and explain what became different for the team or customer. Put methodology beside the claim it qualifies instead of burying every limit in a disclaimer at the end.
A useful results paragraph answers five questions in plain language: What changed? Compared with what? Over which period? According to which source? What else could have influenced the result? That is enough context for most readers and enough specificity for a reviewer to find the underlying evidence card.
Charts should use the same definitions and windows as the prose. Label axes and units, show the baseline, avoid decorative trend lines, and do not crop away an inconvenient period. If the exact values are confidential, use an approved indexed view or describe the operational change instead of manufacturing a vague graph.
W3C's guidance for complex images recommends a short text alternative plus a fuller textual description of the essential information in charts and diagrams. The surrounding article should state the visual's conclusion so the image clarifies the evidence rather than becoming the only place the evidence exists.
Approval needs named owners, not one final thumbs-up
A single “approved” message can hide five different reviews. Route each evidence card to the person who can actually confirm it:
- Scope owner: confirms what was delivered, when, and by whom.
- Data owner: reproduces the metric, definition, filters, and calculation.
- Client approver: authorizes the public company details, quote, role, visuals, and confidential boundaries.
- Editorial owner: checks the net impression, source links, dates, qualifications, and accessible presentation.
- Risk owner: routes regulated, contractual, privacy, name, likeness, or high-stakes claims for appropriate review.
Record the approver, approved wording or asset version, date, and any expiration or reuse condition. Approval to publish one webpage does not automatically authorize a sales deck, paid ad, press release, or future version with stronger wording.
Give the published proof an expiration date
A case study can remain historically accurate while becoming misleading in current sales use. Services change, platforms revise definitions, a client relationship ends, a quoted person changes roles, and an old result may no longer represent the offer being sold.
Set a review date based on risk, not convenience. Recheck active quotes, current role and company naming, links, metric definitions, product or service changes, permission terms, and whether the case study is still being reused in new channels. Preserve the original publication date and historic source record even when the public copy is updated.
If a claim no longer clears review, narrow it, date it more visibly, move it to a historical context, replace it with current evidence, or retire it. Do not quietly swap the period, denominator, attribution model, or quoted wording while keeping the same headline result.
For teams turning complex project work into clear public proof, The Branding Bull's Creative & Branding service covers positioning, message hierarchy, voice, launch assets, and usable file handoff, while Growth Marketing connects measurement, acquisition, and reporting. Send a project brief to discuss the smallest useful case-study, messaging, or measurement scope.
