Most agency portfolios prove the work happened, not that it worked — a wall of thumbnails and a project name. So this page leads with the outcome, names where each figure came from, and says plainly when there isn't a number to show.
Outcome-first cardsSource named per figureGaps disclosedFilter by sector
Also worked with
Government of Alberta
MacEwan University
Canadian Western Bank
AutoCanada
Listed without outcome claims — those engagements' figures aren't ours to publish.
Proof, not promises
Pick the one that looks like your situation.
4 projects · filter to shortlist
SEOAI visibilitySpecialty retail
Six core terms to #1 in nine months.
WC Marker · Alberta
Products people search for by brand, model and country — the kind of query that used to be won purely on rankings. Over the same nine months their referrals from AI answer engines went up four-fold, which is the number that actually mattered: the work that earned the rankings also made them the source the machines quote.
A considered purchase, where the site has to do real explaining before anyone picks up the phone. Discovery concluded rebuild rather than repair: a streamlined user journey, a content structure that could actually describe what they offer, and CRM integration so enquiries stopped being re-keyed by hand.
+164%
Organic click-through lift in year one, from Google Search Console
A city-wide fundraiser — hundreds of kids, one website, and a whole city arriving at once. Event day brought a five-fold surge against the daily average with donations flowing through it the entire time. This is the one day a year when the monitoring and the boring weekly maintenance justify themselves.
People at very different fitness levels arriving with the same question — is this place for someone like me? The redesign simplified that journey rather than adding to it, cutting the route from landing to enquiry down to something a nervous first-timer would actually finish.
No performance figures published for this engagement. We'd rather show a card with an honest gap than fill it with a number we can't attribute — which is the same standard we'd ask you to hold every other agency to.
Twenty years and 280-plus organizations produced a great deal more than four case studies. What's on this page is the subset where the numbers are ours to publish, verifiable, and actually attributable to the work rather than to a coincidence.
The rest fall into three piles: engagements under NDA, engagements where the client's own reporting can't isolate our contribution, and honestly, the ones that didn't move enough to be worth writing up. Every agency has that third pile. Very few mention it.
For context on the selection above: roughly three engagements in ten conclude that a rebuild is the right answer at all — the other seven get optimization work, which produces less quotable case studies and considerably smaller invoices.
a portfolio is a highlight reel — read it that way
Typical agency portfolio
Card headlineproject name
Outcome visible before clickno
Figures attributedrarely
Projects without resultsomitted
Selection bias disclosedno
Proveswork happened
Screenshotsand a client list
This page
Card headlinethe outcome
Outcome visible before clickyes
Figures attributedper figure
Projects without resultsshown anyway
Selection bias disclosedthis section
Proveswork worked
4 of 4sourced or disclosed
Use this on everyone
How to read any agency's case study.
Five tests that separate a real result from a well-designed claim. Run them on this page too — if any card here fails one, that's a fair thing to raise on the call.
01Is the number attributed to a source you could check?
The single fastest tell
"Traffic up 300%" with no source is decoration. Real reporting names the tool and the window — Search Console, GA4, a rank tracker, a specific date range. An unsourced percentage is not evidence, it's typography. If a case study has five impressive figures and no named source for any of them, treat the whole page as marketing copy.
02Percentage or absolute — and why did they choose that one?
Percentages hide small starting points
Going from 4 enquiries a month to 12 is "+200%", which is real but reads far bigger than it is. Absolute numbers hide the opposite problem. Honest reporting gives you both, or gives you the one that's less flattering. When you see only a percentage, ask what the starting number was — the answer is often the actual story.
03Could something other than the agency have caused it?
Correlation gets sold as causation constantly
A seasonal peak, a competitor going under, a funding announcement, a Google update, or the client's own PR campaign. Any of these can move numbers during an engagement. Ask what else was happening in that window — a provider who has thought honestly about attribution will have a considered answer; one who hasn't will look surprised by the question.
04Does the metric connect to money or mission?
Impressions are the classic decoy
Impressions, sessions and keyword counts are easy to grow and often mean nothing. Enquiries, donations, registrations, qualified leads — those are harder to move and harder to fake. If the headline figure is a vanity metric, assume the meaningful ones didn't move, because they'd have led with those instead.
05Where are the projects that didn't work?
The one nobody expects you to ask
Every portfolio is a highlight reel — that's legitimate. What's telling is whether the provider will discuss the other pile when asked directly. Ask for an engagement that underperformed and what they learned. A straight answer is the strongest signal available to you; a deflection tells you how the conversation will go when your own project hits a difficult month.
How to use this: run all five on every shortlisted provider, this one included. A case study that passes 01 and 04 is worth taking seriously. One that fails both is a design exercise. And 05 is worth asking out loud on the call — the reaction tells you more than the answer.
Comparing several agencies
Take this list with you. It's more useful applied to our competitors than to us, and we'd still rather you used it than didn't.
Next step
Does one of these look like you?
Tell me which project on this page resembles your situation and I'll tell you honestly whether yours would go the same way — including the parts of that engagement that were harder than the card makes them sound.
Replies within one business dayYou'll talk to James, not a repHappy to answer question 05
holler