- A credible case study names its timeframe, baseline, metrics and attribution honestly; a persuasive one names whatever survived the cherry-picking.
- The classic tells of fabrication are date ranges that start at a trough, authority scores standing in for traffic, anonymous everything, and timelines no real campaign achieves.
- You do the diligence in ten minutes with four questions, and how an agency answers them tells you more than the case study itself.
Before anyone spends serious money on links, they read case studies, and everyone selling links knows it. The result is a genre: charts that always go up, percentages without denominators, and screenshots cropped precisely where the story would get complicated. Some case studies document real work honestly. The rest are marketing wearing a lab coat.
Telling them apart is a learnable skill, and it takes about ten minutes per study once you know where to look. This guide covers what a credible case study contains, the specific tells of a fabricated or inflated one, and the questions that turn any glossy PDF back into evidence.
Why case studies carry so much weight
Link building is bought on trust. The work happens over months, the results lag the spend, and the deliverables are judged on qualities most buyers cannot fully assess in advance. Case studies fill that gap: they are the closest thing the industry has to a track record you can inspect before signing, which is exactly why they attract embellishment. The incentive to polish is enormous and the audit trail is thin.
None of that makes them worthless. It makes them a document you read the way a lender reads accounts, assuming good faith and verifying anyway. The same posture behind our buyer’s guide to link building services applies here: transparency is the product, and a case study is your first sample of it.
They also do a second job buyers rarely notice: they reveal how an agency thinks. A study built around traffic and revenue shows a team optimising what clients bank; one built around authority-score screenshots shows a team optimising what dashboards display. You are previewing your future monthly report, so read it as one.
Read them in batches when shortlisting, because comparison exposes what a single study hides. Three agencies documenting similar niches with wildly different rigour is the clearest signal a shortlist ever produces, and it costs half an hour.
The anatomy of a credible case study
Honest case studies share a recognisable skeleton, because documenting real work honestly forces the same disclosures every time.
The attribution line deserves emphasis because it is the rarest. Real campaigns coexist with content pushes, technical fixes and algorithm updates, and an honest write-up says so. A study that credits links alone for a 400% traffic rise is describing a controlled experiment nobody actually ran.
Notice what the skeleton does to reading time: with baseline, dates and metrics stated, you can sanity-check a claim in minutes, and the study invites you to. Documents engineered to resist quick checking are communicating something too, and it is rarely confidence.
The tells of a fake or inflated study
Fabrication leaves fingerprints, and the same ones appear across the genre.
- Date ranges that start at a trough. Measuring from a penalty, a migration dip or a seasonal floor manufactures a recovery and sells it as growth.
- Authority scores instead of outcomes. DR and DA can be inflated cheaply; a study bragging about score gains while staying silent on traffic is telling you which one moved.
- Percentages without absolutes. A 300% increase from 50 visits is 150 visits. Studies hide denominators when denominators embarrass them.
- Anonymous everything. Unnamed client, unnamed niche, redacted keywords, cropped screenshots. Confidentiality is real, but a study where nothing is checkable is choosing to be uncheckable.
- Impossible timelines. Competitive keywords to page one in six weeks contradicts how link-driven authority actually moves.
One tell rarely convicts on its own; honest studies sometimes anonymise a client at the client’s insistence. Three tells together is a pattern, and a portfolio where every study shares the same tells is a genre choice rather than a coincidence.
Two more tells operate at portfolio level. Uniform excellence, every client a triumph, no campaign that stalled or taught anything, describes a firm that has never taken a hard client or never reports one, and both readings should worry you. And watch for borrowed proof: aggregate industry statistics or famous-brand logos standing where the agency’s own results should be, which is a portfolio confessing it has none it can show.
Language is the last fingerprint. Honest write-ups use the vocabulary of uncertainty where uncertainty exists, words like contributed, alongside and within, while fabricated ones speak in pure causation because nothing real constrains them.
Reading the charts like an analyst
Most deception lives in chart furniture rather than numbers. Check the axis first: truncated y-axes turn modest gains into hockey sticks, and unlabelled ones turn nothing into anything. Check the metric second: sessions, clicks and impressions are different animals, and a study that switches between them mid-story is picking whichever flatters each slide. Check the window third: a six-month zoom on a two-year chart can hide the decline on either side of the good stretch.
Then run the one verification that is always available: the dates. Cross-reference the claimed growth window against known core updates, and ask what else that site shipped in the period. A gain that coincides with a favourable update is real traffic with shared credit, and an honest agency will volunteer that nuance before you find it.
Screenshots deserve one extra beat of suspicion because tools make them trivially editable. Where a claim matters to your decision, ask to watch the live dashboard on a call rather than accept the image of one; watching a filter get applied in real time is the cheapest audit in the industry, and honest agencies offer it before being asked.
One more chart habit worth stealing from analysts: always ask what is plotted per unit. Links delivered per month is an input; referring domains gained is an outcome; organic sessions to target pages is the outcome that pays. Studies that plot inputs and imply outcomes are counting effort, and effort was never the thing you were buying.
Four questions that make an agency show its working
The fastest diligence is conversational. Send the case study back with four questions and grade the response.
- Can I see the placements? A real campaign has real URLs. Expect a sample rather than the full list, and expect it without drama.
- What was the baseline and what else changed? Honest answers arrive with context attached; evasive ones arrive with adjectives.
- Can I speak to the client, or a comparable one? References exist for work that happened.
- Would this work for my situation, and why? The credible answer engages with your niche and constraints; the alarming one promises the same chart regardless.
Agencies that document honestly enjoy these questions, because they are a chance to demonstrate rigour most competitors cannot match. Discomfort at the first question is common and forgivable; refusal is the answer. Keep a copy of the responses next to the contract, because those written claims about baselines, methods and timelines become the reference points your first quarterly review gets judged against.
If a shortlist survives all four questions, add a fifth on process: ask who exactly would run your account and how placements get approved. Case studies describe the past; the approval workflow describes your future, and mismatches between the two surface here.
Judging results in context
Even a fully honest study needs translating to your situation. A SaaS case says limited things about ecommerce; a campaign run on a strong established domain says little about your young one; a result achieved with an in-house content team firing alongside says nothing about links alone. The right comparison set is businesses at your authority level, in niches of similar difficulty, judged over the timelines in the KPIs guide, and a good agency will help you make that translation rather than resist it.
Sample-size thinking helps too. One spectacular study proves one spectacular outcome happened once; a portfolio of moderate, believable wins across a dozen clients is far stronger evidence about what will happen to you, because it shows a repeatable process rather than a lottery ticket that hit. Weight recency as well: a study from four years ago documents a different algorithm era, and a portfolio where nothing is recent raises its own question.
Finally, ask what happened after the window closed. Durable campaigns show results that held or compounded once the engagement ended; a chart that stops the day the contract did sometimes stops for a reason, and the question costs nothing.
The standard we hold our own studies to
It would be strange to publish this guide and exempt ourselves. Our own case studies follow the skeleton above, and the process behind them is documented for the same reason: real timeframes, absolute numbers with baselines, the work itemised, and credit shared with whatever else the client had in motion. Where a client requires anonymity we say so and keep the rest checkable, and prospects who ask for sample placements get them. That standard costs us the occasional spectacular-looking chart and wins the kind of client who read this far.
Use the same standard on everyone, including us. The agencies worth hiring survive their own case studies being read properly, and the ten minutes this guide costs filters most of the market before a single call, which pairs well with knowing what separates the credible agencies in the first place.
Documentation discipline and delivery discipline travel together, which is the quiet reason this whole exercise predicts outcomes as well as it does.
Want results you could publish as a case study without editing the numbers?
