Case study: 144 → 2,140 organic clicks in a year, with the Search Console report to prove it

Most agency case studies share three flaws: the client is anonymous, the numbers are percentages with no absolutes, and the method is a secret. This one has none of those problems, for a simple reason — the product is ours. EliSpeak is an AI English-speaking tutor we built and market, which means we can put the actual Google Search Console report on the table and walk through exactly what produced it.

Last updated: 9 September 2026. Every figure below is read from Search Console for elispeak.com on this date.

The result, in the report’s own numbers

Comparing the last 3 months (June–September 2026) with the same period a year earlier:

The shape of the 16-month curve matters as much as the totals: roughly twelve months of near-zero — single-digit clicks on most days — and then a takeoff within weeks of the new pages entering the index in May–June 2026. Content SEO does not ramp; it is flat until the moment it is not.

Second-order effect worth noting: the branded query “elispeak” went from 37 clicks to 365 in the same comparison. People who discover a product through a comparison page come back later by name — organic content built brand demand as a by-product.

Which pages did the work

The top content pages for the last 3 months, from the same report:

Every winner is a listicle or an alternatives page. Not the homepage, not feature pages, not how-to tutorials. The lesson generalises: for commercial queries in most niches, Google has already decided it prefers comparison formats — the SERP tells you this before you write a word, if you look.

The six moves, in order

1. Research where the demand actually is — including the traps. Keyword research is less about finding volume than about disqualifying it: the biggest keywords in most niches carry the wrong intent (job seekers, students, DIY hobbyists). The pages above target queries where the searcher is choosing a product, which is why 2,140 clicks from them are worth more than 20,000 clicks of the wrong intent.

2. Copy the format the SERP already rewards. Where listicles rank, write the listicle. Arguing with the SERP — publishing a service page into a results page full of comparisons — is the most common way good content goes nowhere.

3. Build every page to one strict anatomy. A self-contained short answer under the H1 (this is what answer engines quote), published ranking criteria, a comparison table with a “not for” column, per-item verdicts that recommend competitors where they genuinely win, a decision tree, a free plan the reader can run without buying anything, and eight or nine FAQs that are each a real query variant. The honesty devices are not decoration — they are the difference between a page that reads as AI slop and a page that gets cited.

4. Add data nobody can copy. The EliSpeak pages cite aggregates from 1,408 real user onboarding surveys — which learning goals dominate, what learners tried before. A competitor can copy the structure in an afternoon; they cannot copy the dataset. This is the single strongest differentiator available to any product with users.

5. Interlink deliberately. Every content page routes to the two or three money pages it logically feeds, with descriptive anchors. Orphaned pages do not get crawled, let alone ranked — a lesson this site relearned the hard way when its own audit page sat unindexed for months over a missing link.

6. Measure in Search Console, not in vibes. A quarterly year-over-year comparison — the exact report this case study is built on — plus a review loop for pages whose impressions grow but clicks do not. Which brings us to the honest part.

The two numbers that look bad, and what they mean

CTR fell from 2.7% to 0.5%. Composition, not decay: one alternatives page pulled 321,737 impressions — 80% of the site total — on an enormous adjacent query set where it ranks page one but wins a small share of clicks. Impressions grew 75× while clicks grew 15×, so the ratio fell. A small site visible only for its brand name has a beautiful CTR; a site suddenly visible for broad commercial queries has a worse one on vastly more traffic. If a case study you are reading only shows CTR going up, ask what happened to impressions.

One page is doing a fifth of the clicks. Concentration is the norm in content SEO, not a flaw — but it is a risk to name: an algorithm update that reprices one page format can take a visible bite. The mitigation is the system itself: the format is productised, so the next ten pages cost a fraction of the first ten.

What this proves — and what it does not

It proves the system works end to end on a real product in a competitive niche, on a timeline (one flat year, then a steep quarter) that matches what we tell clients to expect. It proves format plus honesty plus proprietary data beats volume publishing — EliSpeak’s four winning pages outperform sites publishing weekly generic articles.

It does not prove the numbers transfer to your niche at the same multiples: a 75× impressions jump owes a lot to one page catching a huge query set, and your market’s SERPs may demand assets you do not have yet. That assessment — whether your niche’s results pages will reward this format — is the first honest conversation of any engagement, and sometimes its answer is no.

We are running the same system on the site you are reading now — the AI automation pages follow the same anatomy, and their Search Console loop runs the same quarterly comparison. When those numbers mature, they will be published the same way this report was.