A commercial insurance broker came to us with a website that described its services well and ranked for almost nothing beyond its own name. It had clients across Australia and offices in a handful of east coast locations, but its site had no page for most of the cities it worked in and only a thin set of pages for the industries it insured.
Over the next seven months we added 10 location pages and about 20 industry pages. Nobody built a backlink to any of them. By the end of that stretch the site was showing up for 5.3 times as many non-brand impressions as before, and the number of non-brand searches where it sat on page one of Google had grown from 194 to 786.
This case study covers what we built, what moved, what didn’t, and the mistakes we’d avoid next time. The business stays anonymous; the numbers come straight out of its Search Console account.
At a glance
- Non-brand impressions per 28-day window: 11,613 in September 2025, 61,468 by April to May 2026.
- Distinct non-brand queries: 841 up to 2,409, and 3,648 by August.
- Non-brand queries ranking in the top 10: 194 up to 786, and 854 by August.
- Four out of every five extra impressions landed on the new location and industry pages.
- The only page that received backlinks in the same period, the homepage, improved the least.
The starting point
In September 2025 the site had around 50 pages that Google showed for non-brand searches. There were service pages for each type of cover, a small blog, and nine older industry pages that averaged position 52 for their queries. That is page six of Google, where almost nobody looks.
Nothing on the site was aimed at a city. Someone in Adelaide or Cairns searching for a business insurance broker had no page on this site written for them, so Google had nothing to show them. For a broker who can service clients anywhere in the country, that is a lot of searches left on the table.
What we built
Ten location pages
Between November 2025 and March 2026 we published one page for each of ten cities and regional centres, from state capitals down to regional towns. The pages shared an outline: why use a local broker, how the broker works with businesses in that area, the cover local businesses tend to need, and a set of questions and answers. The copy under those headings was written for each city. Each page ran to 800 to 1,000 words, and once text that appears on every page of the site is set aside, a typical pair of city pages shared only about 7% of their wording.
That matters because Google’s spam policies call out doorway pages: near-identical pages made for each region purely to catch searches. A set of city pages that differ only by the place name is exactly what those policies describe, and it is the fastest way to waste the effort.
About twenty industry pages
Between late September 2025 and April 2026 we added industry pages for the kinds of businesses the broker already insured: accountants, engineers, farms, construction firms, manufacturers, real estate agencies, life sciences companies and others. Each one was written for its industry, between 500 and 1,200 words, with little wording shared between them.
These pages go after a different kind of search. A business owner rarely types “insurance broker”. They type “insurance for engineers” or “farm insurance broker”, and a site that has no page on the topic has nothing to rank.
What happened
By the April to May window, non-brand impressions had gone from 11,613 to 61,468. They then held there, at 62,440 in May to June and 60,899 in July to August, which tells us the jump wasn’t a temporary spike or a single viral query.
Four in five new impressions came from the new pages
Total non-brand impressions grew by about 49,900 between September and April. Location and industry pages accounted for about 39,900 of that. The homepage, service pages and blog doubled too, from roughly 9,900 to 19,800. Part of that is probably the knock-on effect of a site covering its subject more fully, and part is the blog, which kept growing over the same months. Most of the growth still came from the new pages.
The location pages were uneven, which is normal. The two strongest pages brought in 56% of all location page impressions. The weakest drew 228 impressions in the same window. Regional pages are cheap to write and each one still brings in a steady trickle, so we would build them again.
Location pages showed up for searches with no city in them
The most useful surprise came from searches that didn’t name a place. By July to August, the location pages had appeared 1,104 times in a month for the bare phrase “insurance broker”. In several regional centres they ranked between positions 6 and 9 for it, on page one. In the capital cities the same pages sat much further down, between 14 and 39.
Google localises these searches. When someone types “insurance broker”, Google looks for brokers relevant to where that person is, and a page written for their city is an easy match. The pattern here also shows where the openings are. A regional centre has far fewer brokers competing for that search, so a well-written city page can reach page one there long before it gets near page one in Sydney or Melbourne.
Industry pages climbed from page six to page three
The industry pages averaged position 52 in September 2025. By April to May the average was 22.8, with 178 queries in the top 10 against 22 before. Impressions for this group went from 1,755 to 16,514.
The homepage got the links and barely moved
The broker also received three guest post backlinks in March and April 2026, all pointing at the homepage with the business name as the anchor text. None pointed at a location or industry page.
We compared the homepage with the site’s existing pages that had no new links, using only queries that appeared in both September 2025 and April 2026. The homepage improved by a median of 1.1 positions. The unlinked pages improved by a median of 17.7.
This doesn’t mean links are useless. A homepage already ranking first for the brand name has little room to climb, and in our wider study of 62 websites, a steady flow of links did lift whole sites. It does show where the growth in this case came from: pages that matched what people search for, with no links pointing at them.
What the numbers don’t show
Impressions are not leads. Non-brand clicks rose from 10 to 50 per 28-day window over the same period, a fivefold rise from a very low base. That is real, but it is modest next to the jump in visibility. Only about a quarter of the new pages’ impressions came from page one. Nearly half came from positions 11 to 30, where people rarely scroll, and the rest from further down.
Search Console also hides many rare queries for privacy, so the true click count is higher than the query-level data shows. We still wouldn’t claim more than we can see. The honest summary is that the new pages put the site in front of five times as many searchers and moved hundreds of searches onto page one. Turning that into enquiries is the next stage of work.
The site went through a full rebuild at the end of July 2026. In the first ten days or so after launch, total daily clicks ran 31% above the weeks before and impressions 24% above, so the new pages kept their ground through the move.
What we would do differently
Stop the homepage competing with the city pages
An audit in August found the homepage outranking two of the city pages for searches naming those cities. For one city, the homepage sat around position 10 while the city page was down near 55. Google picked the page with more authority, even though the city page was the better answer. The fix is to link from the homepage and the service pages to each city page using the city in the link text, so Google knows which page belongs to which search. We would do that the day each page goes live.
Write the regional pages further apart
Three of the regional pages ended up sharing between a third and almost half of their wording with each other, far more than the rest of the set. They still picked up impressions, but pages that close together are the ones most at risk of Google treating them as doorways. Next time every city page would get its own local detail from the first draft.
Track keywords from the day a page is published
A rank tracker only watches the keywords it is given, so new pages need their own keywords added on launch day. Search Console shows the results after the fact, but a tracker shows the climb week by week and makes it much easier to see which pages need more work.
Link the new pages into the rest of the site straight away
An August audit found a dozen industry and service pages with no links to them from the body copy of any other page. We fixed most of them that month by linking from related city pages and service pages. Doing it at publication would have helped Google find and understand the pages sooner.
How to apply this to your own site
- List the searches you should appear for but have no page for. Cities you serve, industries you work with, and specific services buried inside a general page.
- Write one real page for each. Different content, different questions and different examples on every page, never a template with a word swapped.
- Start with the biggest markets, then add smaller ones. The large-city pages paid off first, and the regional ones added a steady trickle.
- Link each new page from the homepage or a hub page and from related pages, using wording that names the city or industry.
- Track each page’s keywords from launch and judge it after four to six months, not four weeks.
Questions about this case study
Do location pages still work in 2026?
In this case, yes. Ten location pages went from zero to more than 25,000 non-brand impressions a month within six months of the first one going live, and in regional centres they reached page one for the plain search “insurance broker” because Google localises the results. They only work if each page is genuinely written for its location.
Do location pages need backlinks to rank?
Not in this case. None of the ten location pages had a single backlink, and they still reached an average position of around 24 across hundreds of queries, with 179 of those queries on page one. Links from inside the site mattered more, and we would add them earlier next time.
How many location pages should a business create?
One for each place where you genuinely serve customers and can say something specific. For this broker that meant ten. A business that serves one metro area might need three or four suburb or region pages, and a national business might need dozens. Doorway pages, where the only difference is the place name, are what Google warns against.
How long do new location and industry pages take to rank?
The first pages here went live from late September 2025, and the big jump showed up in the April to May 2026 window, with rankings then holding steady. Expect four to six months before judging a batch of new pages.
