Every agency pitch deck ends the same way. A slide with a green line going up and to the right, a percentage in large type, and a sentence that begins “a personal injury firm in the Southeast.” Partners nod, the marketing lead writes down the number, and nobody asks the one question that would settle whether any of it happened the way it was described.
Case studies are not worthless. They are the only evidence an agency can offer before it has done anything for you, and a good one tells you a great deal about how a firm thinks. The problem is that the format is almost entirely unregulated. There is no auditor, no standard, and no consequence for rounding a story until it stops being true. So the burden of verification sits with the buyer, which means it sits with you.
What follows is how to read one properly. Not with suspicion for its own sake, but with the same discipline you would apply to an expert report from the other side.
The Four Things a Credible Case Study Names
Strip away the design and a case study is a claim about cause and effect. Something was done, and something changed. For that claim to be testable, four elements have to be present. If any one is missing, the study is not evidence, it is a mood.
A Named Client, or a Reference You Can Reach
“A mid-size family law firm in a competitive metro” is not a client, it is a genre. Confidentiality is a real constraint and some clients genuinely forbid attribution, especially firms that would rather competitors not know they use an agency at all. That is fine. What is not fine is anonymity across the board with no path to a live human.
The workable middle ground is this. The agency may anonymize the case study publicly, but at the point where you are seriously considering a contract, it should be able to put you on the phone with the actual client behind at least two of its published studies. An agency that has anonymized every study and can also produce no references is telling you something, and it is not about confidentiality.
Ask the question plainly. “Which of these case studies can I get a reference call on?” The answer arrives fast when the work is real and slowly when it is not.
A Stated Baseline
This is the single most common omission and the most consequential. “Organic traffic up 340 percent” means nothing without the starting number. Three hundred and forty percent on a base of 180 sessions a month is 612 additional sessions, which for a law firm may be a rounding error. The same percentage on a base of 9,000 sessions is a different business entirely.
Percentages without absolutes are the oldest trick in marketing reporting, and they are almost never an accident. A credible study states both. “Organic sessions went from 1,240 per month to 4,050 per month” is a sentence you can reason about. A firm that will not give you the denominator has decided the denominator is unhelpful.
The same applies to conversions. “We tripled their leads” from four to twelve is a real improvement and worth saying out loud in exactly those terms. Dressed up as a percentage without the raw counts, it is being hidden rather than reported.
A Date Range With Both Ends
Two dates, a start and an end, and a note on what was happening before the start. Duration matters because it tells you whether the result is plausible and whether it is durable. A jump inside six weeks on a site that had never been touched usually reflects technical cleanup or a reindexing event rather than sustained content and authority work. A curve that holds across fourteen months is a much stronger claim.
Watch for the range that begins at a convenient trough. If a firm was penalized, migrated badly, or let a site sit broken for a quarter, recovery back to the prior baseline is real work but it is not growth, and presenting the trough as the starting line inflates every number that follows. Ask what the twelve months before the start date looked like. The honest answer is sometimes “they had a bad migration and we fixed it,” which is a perfectly respectable thing to have done and should simply be labeled as such.
Also check the end date. A case study whose data ends nineteen months ago, presented as current proof, raises a fair question about what happened next. Results that continued are usually updated. Results that reversed usually are not.
A Business Outcome, Not a Traffic Outcome
Traffic is an input. Signed cases are the output, and the distance between them is where most law firm marketing money disappears. A case study that peaks at “impressions up 500 percent” has told you about Google, not about the firm. The studies worth reading carry the chain forward, sessions to qualified inquiries to consultations booked to matters signed, and they are honest about where the data gets fuzzy.
Fuzziness is expected. Attribution in legal is genuinely difficult, intake happens by phone, referrals get miscoded, and a client who found the firm through search in March may not call until August. An agency that claims perfect attribution is either using a system you should ask to see or is overstating. An agency that says “we can attribute inquiries reliably and signed matters directionally, here is how we know” is being straight with you. We work through that measurement chain in detail in our piece on measuring law firm SEO ROI in signed cases rather than vanity metrics, and the same logic is what you should be listening for in a pitch.
Screenshots and the Ways They Mislead
Most case studies lean on a screenshot from Google Search Console, Google Analytics, or a rank tracker. These images feel like proof because they come from a system the agency does not control. In practice they are the easiest part of a case study to shape, and usually no fabrication is required at all. Cropping and filtering do the work.
- The truncated axis. A chart whose vertical axis starts at 4,000 instead of zero turns a nine percent rise into a cliff face. Look at the axis labels before you look at the shape of the line. If the labels are cropped out of the image entirely, the shape means nothing.
- The unlabeled property. A Search Console screenshot with the property selector cropped off could be from any site, including the agency’s own. Ask for the version with the property name visible.
- Impressions standing in for clicks. Impressions are the least meaningful line in Search Console and the easiest to move. A site can add tens of thousands of impressions through thin pages that rank on page five for terms nobody with a legal problem is typing. If the chart shows impressions and the caption talks about growth, check whether clicks moved proportionally. Very often they did not.
- The query filter you cannot see. Search Console lets you filter to a query string before screenshotting. Filter to the firm’s brand name and you will capture a beautiful curve that reflects the firm’s billboards, television spend, or a single news cycle. Non brand performance is the only part that reflects SEO work, a distinction worth understanding before any vendor call.
- Average position as a headline. Average position blends every query a page appears for and moves when the mix of queries changes, not only when rankings improve. A site that stops surfacing for irrelevant long tail terms will show a better average position while getting fewer clicks.
- The single keyword ranking. A rank tracker screenshot showing position two for one term proves that one term ranks. It says nothing about the twelve hundred other terms, about whether that term has meaningful volume, or about whether the position holds outside the agency’s tracking location.
- Comparison periods chosen for contrast. Comparing December against the following March flatters almost any legal practice because demand is seasonal. Year over year comparisons are the honest default. Anything else needs a stated reason.
None of this means screenshots are useless. It means you read the axis, the date selector, the property, and the filter row before you read the line. If those are cropped, the correct response is to ask for an uncropped version, and to notice how the agency reacts to being asked.
The Questions to Send in Writing
Reference calls come later. Before you spend anyone’s hour, send five questions by email about the specific case study that impressed you most. Written answers are useful because they are considered, they are on record, and the gaps in them are visible.
- What were the absolute numbers at the start and at the end, not the percentages, for sessions, inquiries, and signed matters if available.
- What else was the firm doing during this period, including paid search, television, radio, billboards, a rebrand, a merger, or a new office.
- What was the monthly investment, and did it change during the period covered.
- What did not work, and what did you stop doing.
- Is this client still with you, and if not, when and why did the engagement end.
Question two matters more than it looks. A firm that launched a seven figure television campaign in month three will see branded search climb sharply, and an SEO agency reporting total organic growth over that window is taking credit for the television. Question four is the character test. Every real engagement has something that failed. An agency that cannot name one has either never examined its own work or has decided you cannot handle the answer, and both are disqualifying. These same instincts apply across the whole selection process, which we lay out in our guide to how to choose a law firm SEO agency.
The Two Reference Calls That Settle It
Written answers narrow the field. Two phone calls close it, and they are deliberately different calls.
The Call With the Featured Client
This is the client from the case study itself. The agency chose them, so expect a friendly conversation. Your job is not to catch anyone out, it is to establish whether the story as told matches the story as lived. Useful questions include what the firm was doing before, what changed inside the firm during the engagement, whether intake staffing or hours changed, whether the reported lead volume felt real at the front desk, and how long it took before anyone noticed a difference.
Ask the operational questions too. Who did they actually talk to each month. Did the same person stay on the account. How much of the firm’s own time did the work require, particularly attorney review time on content, which is the resource partners consistently underestimate. Ask what the reporting looked like and whether they could tell from it what had been done that month.
One question is worth more than the rest. “If you were starting over, would you hire them again, and what would you do differently.” Happy clients answer this with specifics. The specifics are the information.
The Call With a Client Who Is Not in the Case Study
This is the call that actually settles it, and it is the one agencies find harder to arrange. Ask for a reference from a client in a similar practice area and market size to yours who is not featured in any published study, ideally one whose engagement is between eight and eighteen months old, past the honeymoon and long enough to have hit friction.
What you are testing here is consistency. A case study is a best case by construction. The question is what the median client experience looks like, because that is what you are buying. Ask about responsiveness when something went wrong, about how a disagreement over strategy was handled, about whether deliverables arrived on schedule, and about whether the results have been in the same neighborhood as what the agency projected at the start.
If an agency cannot produce this second reference, or offers only more featured clients, treat that as the finding. It usually means the published studies are the outliers rather than the pattern.
What the Case Study Can Never Prove
Even a fully verified case study has a hard limit, and it is worth stating plainly because no agency will state it for you. A result achieved for another firm, in another market, under other competitive conditions, at another point in Google’s history, is not a forecast of your result. It is evidence that the agency has done real work before. That is all it is, and it is genuinely worth something, but it is not a projection and any agency presenting it as one has moved from evidence into sales.
Market difficulty varies enormously. A three attorney estate planning practice in a metro of 400,000 people faces a different problem than a personal injury firm competing in Los Angeles against advertisers spending six figures a month. The same playbook produces different curves. The professional legal technology community has spent years building shared vocabulary and evaluation practice around exactly this problem of assessing vendor claims against a firm’s own circumstances, and the publications library maintained by the International Legal Technology Association is a useful outside reference point for how buying committees at larger firms approach vendor diligence.
There is a related trap on your own site. Firms that scrutinize an agency’s case studies often publish their own results pages with the same weaknesses, no dates, no context, no verifiable specifics. The standards you apply to a vendor apply to your own case results page, where the credibility and the compliance obligations are yours to carry.
A Short Note on Modest Numbers
The most trustworthy case study we have seen in this industry described taking a firm from 61 organic inquiries a year to 143 over twenty months, named the practice area, showed the flat first four months, and noted that two of the six content initiatives produced nothing measurable. No firm would put that on a billboard. It is far more convincing than 340 percent with no baseline, because it has the texture of something that actually happened.
When you review a case study, ask yourself whether it reads like a record or an advertisement. Records contain flat periods, dead ends, and specific dates. Advertisements contain round numbers and adjectives. You have read enough discovery to know the difference on sight, and the same instinct works here.
What to Do With All of This
Pick the two case studies that most influenced your shortlist. Send the five written questions on each. Request the uncropped screenshots. Book the featured client call and insist on the non featured one. Budget about three hours of your marketing lead’s time across two agencies. Against a twelve month contract that will run into six figures at many firms, three hours is not diligence, it is the minimum.
Rubiks publishes case studies with baselines, both dates, absolute numbers, and the parts that did not work, and we will hand you references who are not in them. Talk to Rubiks and bring the competing case study you are trying to evaluate, we will walk you through exactly what it does and does not prove before you sign anything.