Most SEO case studies are marketing collateral wearing a lab coat. A chart goes up, a percentage is bolded, a logo sits at the top, and none of it can be checked. For a Vancouver business trying to decide who to trust with a year of budget, that format is close to useless — it tells you an agency had a good outcome once, but not whether the outcome was caused by the work, whether the starting point was artificially low, or whether anything comparable could happen to you.
This is a template for the other kind of case study: one that shows the work. It is structured so that a sceptical reader can follow the reasoning, identify the gaps, and decide for themselves. If you are an agency, use it to write honest proof. If you are a buyer, use it as a checklist against every case study you are shown — including ours, on the portfolio page.
Start with permission, in writing
Before a single number is published, get written approval from the client for the specific figures, screenshots and claims you intend to use. "They said it was fine on a call" is not permission. Contracts frequently include confidentiality clauses that survive the engagement, and a case study that breaches one is a legal problem, not a marketing asset.
Approval should cover four things separately: use of the company name, use of the logo, disclosure of specific metrics, and any named quote from a person. Clients often approve two of the four. When a client will only approve anonymised disclosure, write the case study as "a Vancouver-based professional services firm with 14 staff" and report percentage change instead of absolute figures. An anonymous case study with real methodology is more credible than a named one with no detail.
Section 1: The business problem, not the SEO problem
Open with what the business was actually trying to fix. Not "organic traffic was flat" — that is a symptom. The business problem sounds like: the company had three sales reps who were relying entirely on referrals, and the referral pipeline had slowed after a competitor opened a second location.
This matters because it establishes whether the case study is relevant to the reader. A commercial landscaping firm reading about an e-commerce brand's traffic growth cannot judge whether the approach transfers. Naming the business problem gives them a way to self-select.
Include: the industry, approximate company size, the market served (Metro Vancouver, BC-wide, national), and what the company was doing for lead generation before the engagement began.
Section 2: The baseline, stated precisely
This is the section that separates a real case study from a sales page. State the starting position with the same precision you use for the result.
- Date range. Name the exact baseline window — for example, the 28 days ending the day before work began. Not "before we started."
- Data source. Google Search Console, GA4, the CRM, the call-tracking platform. Different sources count differently, and a reader needs to know which one produced the number.
- The metrics themselves. Clicks, impressions, average position, sessions, conversions, qualified leads — whichever you will report at the end. Report the same metrics from the same sources on both sides.
- Known data problems. If GA4 was misconfigured, if conversion tracking did not exist, if Search Console had only 90 days of history, say so. Fixing broken measurement is legitimate work, but it also means the "before" number is not trustworthy — and you should be the one to point that out.
A useful discipline: if the baseline is unusually low for a reason unrelated to SEO — a site migration that de-indexed pages, a seasonal trough, a Google Business Profile suspension — disclose it in this section rather than letting a reader discover it. Growth from a broken starting point is real, but it is recovery, not growth, and labelling it correctly costs you nothing.
Section 3: Goals and how success was defined
Write down the goal that was agreed at the start of the engagement, in the words used at the time. Retrofitting a goal to match whichever metric happened to move is the single most common form of case-study dishonesty, and it is nearly invisible to readers.
Good goals are specific and bounded: "increase qualified inbound enquiries from Metro Vancouver by 40% within nine months, measured by form submissions tagged as organic in the CRM." Bad goals are "improve SEO" or "rank higher," because they cannot fail.
If the goal changed mid-engagement — which happens legitimately, when a business pivots or a market shifts — document both goals and the date of the change.
Section 4: Scope of work performed
List what was actually done, in enough detail that another practitioner could evaluate it. This is where most case studies collapse into abstraction ("we implemented a comprehensive technical and content strategy"). Replace that with an inventory:
- Technical fixes shipped, with counts — for example, 340 pages with duplicate title tags corrected, canonical rules standardised, Core Web Vitals work on the two highest-traffic templates.
- Content produced — how many pages, of what type, at what length, targeting what intents.
- Local work — Google Business Profile optimisation, citation cleanup, review acquisition process.
- Links earned or built, and by what method. If you do not want to disclose the method, say that explicitly rather than omitting the category.
- What the client's own team did. Almost every successful engagement involves client-side work — a developer shipping fixes, a subject-matter expert reviewing drafts. Crediting it is honest and makes the case study more useful, because it tells the reader what will be required of them.
Section 5: Timeline
A month-by-month or quarter-by-quarter timeline does two things: it shows the reader how long the work took, and it lets them line up interventions against results. A results chart with no timeline overlay is unfalsifiable.
Include the dates of major shipping milestones and any external events that fall inside the window — a Google core update, a website redesign, a rebrand, the start or stop of paid campaigns. Which brings us to the section most case studies leave out entirely.
Section 6: Confounding factors
SEO does not happen in a sealed room. Before attributing a result to your work, list everything else that could plausibly explain part of it:
- Paid media. Did Google Ads or paid social run during the period? Brand search often rises with paid spend, and brand search inflates organic click totals.
- Algorithm updates. Note any core or spam updates inside the window. A site can gain from an update that penalised its competitors.
- Seasonality. A roofing company measured from February to August will show growth regardless of SEO. Where possible, compare year over year, not period over period.
- Offline and PR activity. A trade show, a news mention, a sponsorship, or a competitor closing.
- Product or pricing changes. Conversion rate improvements are frequently caused by a new pricing page or a shortened form, not by search visibility.
- Reporting changes. A GA4 property rebuild, a switch in attribution model, or new conversion definitions can move numbers without moving the business.
Naming confounders does not weaken a case study. It is the strongest credibility signal available, because agencies with nothing to show avoid the exercise entirely.
Section 7: Results, with definitions attached
Every reported number needs four attributes: the metric, the source, the comparison window, and the definition. "Leads up 62%" means nothing. "Form submissions attributed to organic search in GA4, comparing Mar–Aug 2026 against Mar–Aug 2025, up 62% from 47 to 76" means something and can be argued with.
Some rules that keep results honest:
- Report absolute numbers alongside percentages, or report neither. A jump from 2 to 5 conversions is a 150% increase and is also noise.
- Use year-over-year comparisons for anything seasonal.
- Do not report rankings for a hand-picked list of keywords without saying how the list was chosen. Rankings for terms nobody searches are trivial to produce.
- If a metric went down, report it. A case study where every number improved is either an unusually clean engagement or an edited one.
- Screenshots should show the date range and the property. A cropped chart with no axis labels is decoration.
Section 8: A labelled hypothetical example
The following is a hypothetical illustration. It describes no real client, and the figures are invented purely to demonstrate the format described above. It is included so you can see the structure filled in.
Hypothetical — illustrative only. Not a real client engagement.
Business: A fictional 12-person commercial cleaning company serving Metro Vancouver, historically dependent on a single property-management referral partner.
Baseline (28 days pre-engagement, Search Console + GA4): 210 organic clicks, 18,400 impressions, average position 42.6, 4 organic form submissions. GA4 conversion tracking was firing on page view rather than form submit, so the pre-period conversion figure is unreliable and is reported here only for completeness.
Goal: Reduce dependence on one referral source by generating 20+ qualified organic enquiries per month within 12 months.
Scope: Conversion tracking rebuilt in month 1; 22 duplicate service pages consolidated; 9 new service-by-area pages published; Google Business Profile categories and services corrected; 40 citations cleaned; internal linking restructured around four service hubs.
Timeline: Months 1–2 measurement and technical remediation. Months 3–7 content. Months 8–12 local signals and refinement.
Confounders: A Google core update landed in month 6. The client ran a $1,200/month Google Ads brand campaign throughout, which inflates branded organic click volume. A new pricing page shipped in month 9 and likely contributed to conversion-rate change independently of search.
Result (months 10–12 vs. same months prior year, Search Console + GA4): organic clicks 210 → 690 per 28 days; average position 42.6 → 19.4; organic form submissions 4 → 23 per month, measured with corrected tracking. Because the baseline conversion tracking was faulty, the conversion comparison should be read as directional rather than exact.
What did not work: Two of the nine area pages never gained impressions and were consolidated in month 11.
Notice how much weaker — and how much more believable — that reads than "we grew organic leads 475%."
Section 9: Claim-integrity checks before publishing
Run every case study through this list before it goes live:
- Can every number be traced back to a screenshot or an export you still hold?
- Is the comparison window the same length on both sides?
- Have you avoided implying causation where you can only show correlation? Prefer "during this period" to "as a result of."
- Is any client-identifying detail present that the client did not approve?
- Are testimonials real, attributed and approved? Do not publish review or rating schema for a quote you assembled yourself.
- Would the client's own analyst read this and agree with it?
- Is the hypothetical labelled at the top of the block, not in a footnote?
On schema specifically: mark up a case study as an Article with a BreadcrumbList. Do not add Review, AggregateRating or Product markup to manufacture a rich result. Self-serving review markup is against Google's structured data guidelines and is one of the easier things for a competitor to report.
Section 10: What buyers should ask
If you are on the buying side in Vancouver and an agency hands you a case study, these five questions will tell you almost everything:
- What was the baseline, and what date range does it cover?
- What else was running during the period — paid media, PR, a redesign?
- Which of these results came from work your team did versus the client's team?
- What is in this engagement that did not work?
- Will the named client take a 15-minute reference call?
An agency that documents its work answers all five in a single conversation. An agency that cannot answer the first one is showing you a chart, not a case study. The same logic applies to the questions we recommend asking any provider in our guide on choosing a Vancouver SEO company.
Using the template
Copy the ten sections above into a document and fill them in as the engagement runs, rather than reconstructing them a year later from memory. Capturing the baseline on day one takes twenty minutes and is impossible to recover afterwards. Logging confounders monthly is a two-line note. The case study then writes itself, and it will be true.
If you want to see how we document our own work, our case studies are here, and the scope of what a full engagement covers is set out on our SEO services page. If you would rather start with your own numbers, the free SEO audit tool will give you a baseline snapshot of where your site stands today — which is exactly the input Section 2 needs.
