Local SEO Case Study: Two Properties, 16 Months, Every Export Attached

The local SEO case study genre has a receipts problem

Almost every local SEO case study published is the same artefact: a Google Maps screenshot, a percentage, a 90-day claim, and a client who can’t be named. There’s no export, no window, no query list, and no way for you to tell whether the ranking moved because of the work or because a competitor’s site went down.

The opponent here is that format, not local SEO.

I own the two properties below, so the exports can be published in full: the query list, the position bands, the device split, the dated screenshots. Both windows close on 12 August 2026 and both figures came out of Google Search Console on the same day.

Both properties found the same thing, from opposite directions

The city guide and the pilgrimage guide don’t have much in common. One publishes bus timings and hospital prices continuously; the other answers a question three million people ask over eleven days in September.

They arrived at the same finding anyway.

Position stopped predicting the outcome. CoimbatoreJunction.in averages position 10 and clicks at 0.98%. VelankanniShrine.in averages position 9.1 and clicks at 3.19%. Same country, same tool, same window, near-identical average position, and a click rate three times apart.

The variable isn’t where the result sits. It’s what the searcher was trying to do when they typed it.

Velankanni’s queries are tasks. Book a room, find the flag hoisting date, get the mass schedule, work out the train. Somebody with a task in front of them will scroll past three results to find the one that completes it, which is why the booking office page sits at position 7.34 and converts at 9.27%.

Coimbatore’s queries are lookups. A pincode, a ward number, a gold rate. Those get answered in the SERP, and increasingly by a model reading the page rather than a person visiting it.

The part I did not expect

I built the city guide on the assumption that out-answering the municipal corporation would produce traffic, because I was never going to out-authority it.

The answers worked. The traffic didn’t follow.

Clicks fell over the window while Copilot citations quadrupled to 40,101 in 180 days, and the single most-cited page on the property converts at 0.09%. Two pages on that site have near-identical organic impression volume and opposite outcomes: the ward list is a short lookup people click through to, and the red taxi guide is a 47-question reference with real pricing tables that the SERP and the models can answer from without sending anybody anywhere.

That’s the uncomfortable version of a local SEO win. The structure that made the page the answer is the same structure that removed the reason to visit it.

What actually moved the numbers

Four decisions carried both properties, and none of them is a trick. If you’re running a local site, these are the four worth copying.

Pages come from queries, not from a content calendar. Every page on the city guide exists because people were already searching for that specific thing: a ward number, a hospital package price, a snake catcher. Nothing was commissioned because it seemed like a good topic.

Real HTML tables, never images of tables. Prices, routes and timings are marked up as tables. An image of a table is invisible to a crawler and useless to a model, and the pricing pages using real tables are the ones quoted verbatim. I can’t rule out other factors, but the pattern has held on every property where it was tested.

Freshness as a schedule with dates on it. A bus timing that was right in March is wrong by September, and a wrong local answer is worse than no answer. Once your page is the cited answer, staleness stops being an inconvenience and becomes a liability you own.

Internal links follow the task, not the taxonomy. Somebody who looks up a bus route needs the timings next, then the fare. Both properties link in that sequence rather than to a related category, which is also why the pilgrimage guide has 57 individual accommodation pages cited by Copilot rather than one overview page taking all of it.

What to take from this if you run a local site

Measure the citation, not only the click.

If your pages answer factual local questions properly, some of your best work won’t show up as traffic at all. It will show up as a citation count in Bing Webmaster Tools sitting next to a flat line in Google Analytics, and you’ll conclude the work failed.

That conclusion is the real risk here, and it’s the one I nearly reached myself on a property doing 3.9 million impressions. The AI SEO version of this evidence sets out the reports to check instead.

The second thing worth copying is cheaper than it sounds. Mark your prices, timings and routes up as real HTML tables, and if you’re publishing them as images today, that’s the highest-return afternoon of work available to you.