Skip to content
Case study 01 · blazehive.io · Jan – Jul 2026 · Source: Google Search Console

From 25 clicks a month to 212 a day.

This is our own domain. We pointed the BlazeHive engine at blazehive.io and let it run. Every number on this page is pulled from the Google Search Console API on the domain property, and we will open the live account on a call so you can drive.

25
clicks, all of January
212
clicks on Jul 21 alone
74 → 13.2
avg. position, Jan → Jul
645,030
impressions, last 28 days
4,000+
distinct ranking queries
340+
URLs earning impressions
0 50 100 150 200 JANFEBMARAPRMAYJUNJUL REBUILD TOOLS + CALC 174 LANDING Q-LAYER 212 clicks · Jul 21
Organic clicks per day, Jan 1 – Jul 21 2026 · thin line = daily · thick line = 7-day average · dashes = ship dates
Jan 2026

Where it started

blazehive.io was not a fresh domain in January. It had pages live, and Google saw them: about 5,000 impressions a day. It just refused to rank them anywhere a human would look. Average position was 74. That is page seven of the results.

The whole month produced 25 clicks. So the starting point was worse than zero pages: it was pages Google had already seen and decided to bury.

Apr – Jul 2026

What we shipped

Between April and July we rebuilt the site and published in layers. Each layer is a page type. Each page targets one keyword cluster. Every page goes through the same pipeline: keyword scoring, cannibalization check, drafting, validation, publish.

Date Layer Job What it is
Apr 05 Site rebuild Foundation New architecture, static rendering, every page two clicks from the homepage.
Apr 22 Free tools Acquisition Working utilities people actually use, not articles about utilities.
Apr 28 Calculators Acquisition Same idea, numeric niches. One of these becomes the stress test below.
May 11 Comparisons, alternatives, listicles AI visibility The page formats AI assistants quote when someone asks them what to buy.
May 29 174 landing pages Conversion One bottom-of-funnel keyword cluster per page, one shared template.
Jul 13 80-page question layer Authority Built from a mined question graph. More on this below, but not much more.

Every layer has one job. Most content programs treat everything as "the blog" and measure output in posts per month. We treat page types like positions in a portfolio, each with its own return profile:

Landing pages convert. They target queries where the searcher already knows what they want and is choosing a provider. The page does not educate; it is the thing being searched for. Traffic on these pages is small next to the tool pages, but it is the traffic that fills pipelines, which is why the layer gets the most engineering per page.

Question pages build authority. Google reserves its best rankings for sites that demonstrably cover a topic in depth, not just its commercial center. Each page in the July layer answers one real question searchers ask around our revenue keywords, and each one channels its authority to exactly one revenue page. The layer exists to make the landing pages rank harder, not to collect traffic of its own.

Comparisons, alternatives, and listicles earn AI visibility. When someone asks ChatGPT or Perplexity for "the best X" or "alternatives to Y", the assistant builds its answer from exactly this page format. A growing share of buying research now starts in an AI chat instead of a search box, and no analytics tool shows you that traffic until the customer arrives already convinced. These pages are how a brand gets into those answers, and they are the youngest bet on the site: one alternatives cluster is already collecting 9,700+ monthly impressions while its positions climb.

Free tools and calculators pull volume. A working utility earns clicks, bookmarks, and repeat visits that no article ever will. They bring the audience; the other layers monetize the attention.

Underneath all of it, the filter does most of the work. We score every candidate keyword with a proprietary intent model and throw away the large majority. What survives is one specific kind of query: the searcher already knows what they want, and a page can be the answer rather than an article about the answer. No "what is", no "tips", no glossary posts. Volume barely matters at this stage; some of our best pages target keywords a volume tool reports at 10 searches a month.

No two pages compete with each other. Before a page ships, its keyword cluster runs through our own cannibalization detection algorithm, which checks it against every page already live. If two pages would chase the same query, they get merged, re-scoped, or one gets killed. This is the unglamorous reason the curve at the top is smooth: sites that skip this step grow in bursts, then flatline while their own pages fight each other for the same rankings.

One page carries a whole cluster. Each page targets one exact keyword plus its variants, placed in a fixed set of on-page positions (the title, URL, top heading, and opening line among them). We deliberately stop short of a perfect on-page score. Pages that optimize to 100 look like pages that optimize to 100.

Nothing ships unvalidated. Every page passes automated gates before publish: keyword placement, structure, cross-page repetition, and a lint pass against a catalog of AI writing patterns. Pages that fail get fixed or killed.

The July layer is the one people ask about. We mined 263 real questions searchers ask around our revenue keywords, collapsed duplicates and intent variants down to 80, and published each one as its own page, wired so every question page feeds authority to exactly one revenue page. Two weeks old at the time of writing and already the fastest-indexing layer on the site. That is as much as we put in writing.

Receipts

Position moves, query by query

Averages hide things, so here are individual keywords. April versus the last 28 days:

Query Pos. Apr Pos. now Impr. 28d
backlink checker 56.3 11.1 80,597
free backlink checker 13.5 7,360
link building services 9.5 9,174
mobile home value calculator 6.5 1.4 525
used mobile home value calculator 10.4 1.5 381
how much is my mobile home worth 12.3 1.9 865

Positions are Search Console 28-day averages (Apr 1–28 vs Jun 24 – Jul 21). — means the query had no impressions in April.

Two of those are worth a sentence each. "backlink checker" is a head term the biggest SEO brands compete for; our page moved 45 positions in twelve weeks and now collects 80,597 impressions a month from that one query. And "link building services" is a pure commercial term where page one is worth actual money; we sit at 9.5 and climbing.

Control

The off-niche stress test

A fair objection at this point: of course an SEO company can rank an SEO site. It is our home field.

So here is the control. Among the calculator layer we ran a page in a market where we have no authority, no history, and no domain relevance at all: mobile home valuation. In January that page sat at position 36.9 and earned 2 clicks. We rebuilt it through the same pipeline in April.

Last 30 days: 3,624 clicks, average position 3.8, and position 1–3 across 30+ query variations, from "mobile home value calculator" (33.9% CTR at position 1.4) to "used mobile home value by serial number".

That is the point of the test. The method does not depend on us knowing your industry. If it ranks in a niche we know nothing about, it ranks in yours.

Jul 2026

Still compounding

1 20 40 60 80 JANFEBMARAPRMAYJUNJUL pos 13.2
Average position across all queries, daily · down is good · 74 in January, 13.2 now

Monthly clicks, January through July:

25
Jan
28
Feb
65
Mar
385
Apr
1,281
May
2,896
Jun
3,270
Jul·21d

Not one month went backward. Google shipped its usual cadence of algorithm updates inside this window, and the curve absorbed all of them without a drawdown, because pages built for buying intent are what updates reward, not what they correct. Sites built on thin informational content live in fear of update season. This one grew through it.

Impressions are the leading indicator, and they are accelerating: 11,709 on the first day of the last 28-day window, 41,777 on the peak day at the end of it. Average position is still falling, and the question layer from July 13 is barely indexed. The click curve at the top of this page is not a plateau; it is the middle of the ramp.

Verify

Why you can trust these numbers

None of this is a screenshot. Every figure comes from the Search Console API on sc-domain:blazehive.io, dated, and reproducible. On a call, we open the live property and you drive.

And remember whose site this is. blazehive.io is our own product. We are the client with the most to lose if the method stops working, which is exactly why we run it on ourselves first.

Your domain next

The same engine, pointed at your site.

Drop your URL and BlazeHive runs the identical pipeline: keyword scoring, page layers, validation, one ranked page every morning.

BlazeHive · Case study 01 Data window: 2026-01-01 – 2026-07-21