Case studies are marketing documents. That does not make them useless, it means they need reading in a particular way.

The most informative parts are usually the ones nobody highlighted: the timeline, the starting problem, and the figures that seem to disagree with each other.

Here is how to read one properly, using a decade long architecture engagement as the example, and what it tells you about how seo for architect firms actually behaves over time.

Read The Client Start Date Before Anything Else

The single most useful line in any case study, and it is rarely in the headline.

This London practice has been a client since early 2016. Every figure in the case study represents roughly a decade of continuous work, not a campaign.

That reframes everything. An authority score of 31, 1.5K referring domains and 5.2K backlinks are not quarterly outcomes. They are what a decade of compounding produces, and a practice reading those numbers should adjust its own expectations accordingly rather than asking why six months has not delivered them.

Check this on every case study you read. Absence of a start date is itself informative.

Look For The Original Problem, Not The Achievement

The problem statement tells you whether the case resembles your situation.

Here it was flat impressions, organic visitors barely moving month to month, almost no presence in AI answers, and inconsistent enquiries causing the pipeline to dry up between projects.

That last item is the one architecture principals will recognise. Feast and famine between projects is the actual business problem, and it is more specific than a traffic complaint. A case study that names a business problem rather than a metric problem was written by someone who understood the client.

If the stated problem is only ever low traffic, the engagement was probably scoped as a traffic exercise.

Treat Conflicting Figures As Information Rather Than Deception

Worth being straightforward about this, since it affects how you read every case study including this one.

The Extension Architecture page carries more than one traffic figure. Its meta description cites 11.6K monthly visits and 8K ranking keywords, while the results section states 10.6K organic traffic and 6.1K organic keywords. Elsewhere the page references 15.9K for the UK location specifically.

These are almost certainly different measurement scopes and different snapshot dates rather than contradictions, since organic figures move monthly and pages get updated at different times. Using the block explicitly labelled as current is the safest reading, which gives 10.6K organic traffic and 6.1K keywords.

The transferable lesson is not about this agency. It is that any percentage or figure in any case study needs three questions attached: measured when, across what scope, and against what baseline. Ask them every time.

The First Ninety Days Reveal The Real Sequence

Where a case study documents its early months, read that section first.

Month one here was technical: crawl errors fixed, site speed addressed, schema rolled out across every page so search engines could read the content properly. Month two was content and structure, service and project pages mapped to real demand with clear internal linking. Month three began authority work, outreach and AI citation optimisation, with early keyword movement starting to show.

Nothing was published in month one. That order is the finding, and it is consistent with how this discipline works regardless of sector. Practices that begin with content on an unreadable site produce activity without movement.

Notice Which Channel Carried The Early Period

The detail most retellings omit, and the practical one.

Paid campaigns ran on high intent extension and conversion searches specifically to bring leads in while organic rankings climbed, and those campaigns produced 4x growth within the first three months.

So the organic work had three months of cover. A practice without that bridge spends the same period with nothing arriving, which is exactly when engagements get cancelled.

Read case studies for this. If paid ran alongside and the case study only reports organic gains, the timeline you are being shown was not experienced as a dry spell.

Check Which Queries The AI Citations Came From

Citation counts mean less than the queries producing them.

The prompts here are the ones a homeowner would genuinely use: best architect for a house extension, house extension architect in London, planning permission extension cost, architect near me for an extension, London extension architect reviews.

Every one sits at decision stage rather than research stage. That distinction determines whether citations translate into enquiries, and a case study listing its actual queries is telling you something a total figure cannot.

One caution on this particular page. Its AI mention counts appear inconsistent between sections, so the page count of 148 cited pages is the figure I would rely on rather than the mention totals.

Distinguish Continuous Work From A Project

The achievements section here states that continuous work across AI visibility, GEO, organic search, paid and content has improved things gradually, month over month.

Gradually and month over month are the operative words. This is not a case of a fix producing a jump. It is accumulation, which for a practice deciding whether to commit budget is the more honest picture and the less exciting one.

Case studies describing a sudden transformation are either reporting a single technical unblocking, which is real but rare, or compressing a long timeline for narrative effect.

Ask What The Practice Had To Supply

Almost never stated and worth inferring.

Schema across every service and project page, content mapped to extension and conversion intent, project pages published, and digital PR outreach to architecture and home improvement publications. None of that happens without the practice supplying project detail, approving copy, and being available for outreach.

A decade of that requires someone internally who kept answering. That is the invisible half of every case study and the part most likely to explain why your engagement went differently from theirs.

What This Case Genuinely Supports And What It Does Not

Fair separation, since this is the point of reading critically.

Supported: that sustained technical, content, authority and AI work over years produces substantial compounding results for an architecture practice. That the sequence should be technical, then content, then authority. That paid can cover the early period.

Not supported: any timeline expectation for a new engagement. Ten years of results tell you the ceiling, not the pace.

Not supported: that these figures are typical. Big Hunt states three or more architecture studios served, and its other published London practice shows considerably more modest numbers.

Ask For The Ordinary Account, Not This One

The practical instruction that comes out of all of the above.

Any agency will show you its best case study. Ask instead for a client in your sector whose results were slower than expected, what caused it, and what changed. That answer tells you more about competence than a decade of compounding does, because it reveals how they behave when the plan is not working.

Then ask for the start date on every figure they show you.