Per Scrunch’s own tracking (read September 7, 2026), homecalc.ca is the single most-cited domain among its tracked competitors: 63 citations against 45 for the next domain. But on mentions(the AI actually naming “HomeCalc” in the answer), it trails Ratehub and WOWA. And despite real citation and mention activity, HomeCalc has recorded zero AI-referred sessions on Google Analytics over the last 90 days. Three honest, separately-sourced findings, not one tidy story.
Disclosure. HomeCalc.ca is a site I own; it is the source of the case study linked throughout this site. I have no commercial relationship with Scrunch: this is not a sponsored review, and I paid for the account like any other customer. Every number below was read directly from the live dashboard on September 7, 2026 and is labeled with which view it came from, because two views on the same dashboard showed slightly different response totals for the same period (see the note in the methodology section).
HomeCalc.ca already has two other measurement instruments on this site: a live Bing Webmaster Tools citation count, and a published, DOI-registered dataset from a manually-run prompt panel. This is a third and separate one. Scrunch runs its own prompt set against ChatGPT, Perplexity, Google AI Overviews, Microsoft Copiloton a schedule and tracks HomeCalc against five named competitors. The numbers here don’t replace the other two; they add a different lens, and they are kept clearly labeled so nothing gets blended.
01: What was measured
Scrunch tracked 20 prompt variants (from 5 seed prompts, against a 500-variant plan cap) across four engines: ChatGPT, Perplexity, Google AI Overviews, Microsoft Copilot. Five named competitors are configured for comparison: myperch.io, nesto.ca, Ratehub (ratehub.ca), rates.ca, WOWA (wowa.ca).
Two different views inside the same dashboard, read the same day, showed different response totals for what looked like the same window: the Home dashboard reported 140 total responses (Aug 30 - Sep 6, 2026 (Home dashboard default view)), while the Explorer > Citations view reported 110 total responses for Sep 1 - Sep 6, 2026 (Explorer / Brands page window). The headline rates (mention rate, citation rate, sentiment) match across both; only the underlying counts differ by the exact window each view defaulted to. Both are cited below, labeled by which view they came from, rather than picked to make one number look bigger.
Headline numbers, per the Home dashboard: a mentions rate of 17.1%, a citations rate of 45%, and a sentiment average of 100 (100% of scored responses came back positive). On position, when HomeCalc appeared, 62.5% of the time it held a top position and 37.5% a middle one.
02: Top of the citation table, not the mention table
On raw citation count, per the Explorer > Citations view (Sep 1 - Sep 6, 2026 (Explorer / Brands page window)), homecalc.ca leads every tracked domain:
| Domain | Citations |
|---|---|
| homecalc.ca | 63 |
| wowa.ca | 45 |
| cmhc-schl.gc.ca | 43 |
| ratehub.ca | 40 |
But per the Home dashboard’s Top Brands table (Aug 30 - Sep 6, 2026 (Home dashboard default view)), on mentions rate, HomeCalc trails two of the same competitors:
| Brand | Mentions rate |
|---|---|
| HomeCalc | 17.1% |
| Ratehub | 24.3% |
| WOWA | 21.4% |
| rates.ca | 3.6% |
Same tool, same rough window, two different leaderboards. The domain-level table above counts every response where a HomeCalc URL was used as a source. The mentions table counts only responses where the AI’s own text names the brand. HomeCalc wins the first and loses the second to Ratehub and WOWA.
03: The land-transfer-tax page: cited, never named
The clearest single example sits on one topic. Scrunch’s Citation Mention Rate column shows, for the same broad query set, how often a domain’s citation comes with that brand actually named in the answer:
| Domain | Citations rate | Named when cited? |
|---|---|---|
| homecalc.ca | 45% | 100% (HomeCalc) |
| wowa.ca | 32.1% | 100% (wowa) |
| ratehub.ca | 28.6% | 100% (ratehub) |
| nesto.ca | 11.4% | 100% (nesto) |
Narrowed to land transfer tax specifically, the same pattern holds at the page level:
| Page | Citations | Named when cited? |
|---|---|---|
| homecalc.ca/tools/land-transfer-tax-calculator | 7 | 100% (HomeCalc) |
| wowa.ca/calculators/land-transfer-tax | 14 | 100% (competitor) |
| ratehub.ca/land-transfer-tax | 13 | 100% (competitor) |
WOWA’s and Ratehub’s land-transfer-tax calculator pages get cited and named together, every time, in this dataset. HomeCalc’s equivalent page gets cited and never named in the same breath. Nobody outside the model providers can say exactly why with certainty; the pattern is consistent with an entity-clarity gap on that specific page (the calculator doing the math without stating clearly enough whose tool it is) rather than a ranking or technical problem, given what shows up in the audit next.
04: A clean technical audit, and 22 of 24 pages with zero citations
Scrunch’s Site Maps tool crawled 24 HomeCalc pages (of a 25-page plan limit), last crawl 2026-08-31. Every single one scored 100, “Excellent”on Scrunch’s own page-quality audit. That rules out the technical layer as the explanation for what comes next: only 2 of the 24 tracked pages had any citations at all in this window, the homepage (12) and the house-flip tax calculator (4). The other 22 pages, including well-optimized guides on rent-vs-buy, capital gains tax and mortgage qualification, scored perfectly on structure and still had zero recorded citations.
The lesson isn’t that the audit tool is wrong; it’s that a perfect technical score is necessary and nowhere near sufficient. AI engines are pulling almost exclusively from HomeCalc’s calculator tools, not its blog content, even though the blog content passes every structural check Scrunch runs.
05: Citations without clicks
Scrunch’s AI Referrals tab reads directly from a connected Google Analytics property (integration status: Active, confirmed, not a missing-integration false zero). Over the last 90 days: 0 sessions, 0 views originating from an AI platform.
That sits next to real citation and mention activity above. Being used as a source, and even being named as a brand, is not the same as someone clicking through and landing on the site. This is the gap between AI visibility and AI-referred traffic, and right now for HomeCalc it’s a real gap, not a rounding error.
06: A caution about the tool itself
Scrunch’s Configure section auto-suggests sub-brands it detects in response text, with a confidence score. 10 suggestions were pending review at read time. Cross-checking two of them against the Citations domain table turned up a real problem:
| Suggested as | Confidence | Actually is |
|---|---|---|
| Maple Syrup Money | 88-89% | maplesyrupmoney.com (13 independent citations) |
| CanCalc | 88% | cancalc.com (8 independent citations) |
“Maple Syrup Money” and “CanCalc” are real, independent domains competing for the same prompts, not HomeCalc sub-brands. Accepting either suggestion would have folded a competitor’s citations into HomeCalc’s own numbers. Neither was accepted. The broader point: an AI-detection feature inside an AI-tracking tool is still worth checking by hand before you trust it, the same discipline this site applies to every number it publishes about itself.
07: What this isn't claiming
- Not a claim that HomeCalc outranks Ratehub or WOWA overall: it leads on citation share and trails on mention share, in the same dataset.
- Not proof of why the land-transfer-tax page gets cited without being named: that's a pattern worth testing against, not an established cause.
- Not evidence Scrunch is inaccurate: the misattributed suggestions were caught precisely because Scrunch surfaced them for review rather than auto-applying them.
- Not merged with the Bing Webmaster Tools citation count or the published Zenodo dataset elsewhere on this site: three different instruments, three different numbers, on purpose.
08: Frequently asked questions
Is this the same HomeCalc data as your other case study?
No, and that's deliberate. The existing HomeCalc case study and the citation-study dataset come from Bing Webmaster Tools and a separately published prompt panel. This post is a third, independent instrument: Scrunch's own tracking tool, running its own prompt set against ChatGPT, Perplexity, Google AI Overviews and Microsoft Copilot over a much shorter window (early September 2026). All three measure the same site and generally point the same direction, but the numbers are not interchangeable and should never be added together.
What's the difference between a citation and a mention in this data?
A citation means HomeCalc's own domain appears as a linked source in the AI's answer. A mention means the AI's written text actually names "HomeCalc" as a brand. Scrunch tracks these separately, and the gap between them is the finding this post is built around: HomeCalc's pages get used as sources without the brand being named in the same breath, while several competitor pages on the identical topic get both.
Does a perfect technical audit score mean nothing was wrong?
It means the technical layer Scrunch checks (page structure, accessibility, crawlability) wasn't the problem here. All 24 tracked HomeCalc pages scored 100/Excellent, and 22 of those 24 still had zero citations in the tracked window. A page can be technically flawless and still not be the one an AI engine chooses to pull from or name; that's a content and entity-signal question, not a markup one.
Why would a citation not come with a brand mention?
Nobody outside the model providers knows the exact mechanism, and this post doesn't claim to. What's checkable is the pattern: on the land-transfer-tax topic, HomeCalc's own calculator page gets cited but never named in the same answer, while WOWA's and Ratehub's equivalent pages get cited and named together every time in this dataset. That's consistent with an entity-clarity gap (the page doing the math without stating clearly enough whose tool it is) rather than a ranking problem, but it's a hypothesis worth testing, not a proven cause.
If citations aren't producing clicks, what's the point of tracking them?
Citations are a leading indicator, not the outcome itself. In this dataset, HomeCalc had zero AI-referred sessions in Google Analytics over 90 days despite real citation and mention activity. That doesn't make the citation data worthless: it tells you the site is present in the answers but not yet winning the click, which is a different, and arguably harder, problem than not being present at all.
This kind of citation-vs-mention gap, and the technical audit-vs-content gap next to it, is exactly what an AI visibility audit is built to find on your own site, across six platforms, not just the four Scrunch tracks here. See the full HomeCalc case study or the site’s measurement methodology for how these numbers are kept honest.