Transparent methodology.

Here is exactly what we ask, where we ask it, how your grade is calculated — and what the method can't tell you.

Step 1 · The questions

We ask what your customers ask

6 question types per trade, written the way real people type them, with your city substituted into every one. Each covers a different intent — from research to "my-basement-is-flooding". Here's the set for a Toronto plumber:

  • 1"best plumber in Toronto"
  • 2"who is the most reliable plumber in Toronto"
  • 3"top rated plumber in Toronto 2026"
  • 4"I need a plumber in Toronto, who should I call"
  • 5"plumber in Toronto with good reviews"
  • 6"burst pipe repair Toronto"

Query set v1 — sets are versioned and append-only, so scans stay comparable over time.

Step 2 · The engines

Every question, three times, on four engines

AI answers vary between runs, so one answer proves nothing. Every question runs 3 times on each engine with live web access: 6 questions × 4 engines × 3 runs = 72 answers analyzed. The repeats measure consistency — they are not 72 different questions.

ChatGPT

web search on · 18 answers

Claude

web search on · 18 answers

Gemini

Google grounding · 18 answers

Perplexity

native web index · 18 answers

Step 3 · Reading the answers

How answers become data

Each answer is read by an extraction model that records: whether your business was named, in what position, how it was described (positive, neutral, or negative), and every other business the answer recommended.

Names are matched loosely — capitalization, punctuation, and small variations like "Ltd." don't break the match. Franchises that share a brand name across cities are matched within your city's answers, which is one reason the scan form asks you to confirm the exact business name.

Step 4 · The grade

Scored out of 100, graded A–F

How often you're named

Your mention rate across all 72 answers — the core of the grade.

60 pts

How high you appear

Named first earns 20 points, second 12, third 6 — averaged over the answers that name you. Callers rarely reach the third name.

20 pts

How you're described

Recommended positively (10), listed neutrally (5), or flagged negatively (0) — averaged over your mentions.

10 pts

How many engines know you

Full marks requires at least one mention on each of the four engines.

10 pts
A80–100Named consistently across the tested questions. Defend it.
B65–79Named often, but rarely first. Close the gap.
C50–64Visible but inconsistent — most answers still miss you.
D35–49Mostly invisible, with occasional mentions on one engine.
F0–34The engines didn't surface you for the tested questions.

The honest part

What this method can't tell you

  • AI answers change over time and can vary between identical runs — a scan is a snapshot of the scan date, not a guarantee of future answers.
  • The 6 questions per trade are a representative subset of how locals ask — not every possible phrasing, neighbourhood, or service niche.
  • Business names are matched by an extraction model with loose normalization. Franchises sharing one brand name, recent renames, and unusual spellings can be miscounted — the confirm step in the scan form exists to reduce this.
  • We measure what public engines answer. We have no access to any engine's internal ranking systems, and no way to pay for placement — nobody does.
  • A grade measures recommendation visibility on the tested questions. It does not measure workmanship, and a higher grade is not a guarantee of more leads.

Privacy: we never send your email or phone number to AI engines — scans use public business information only. LocalRankHQ is an independent service and is not affiliated with or endorsed by OpenAI, Anthropic, Google, or Perplexity.

Want to see the output? View a sample report →

Find out who AI recommends before your next customer does

When customers ask AI who to hire, we show whether your business is in the answer.

Scan My BusinessHow it works

Free · No credit card · Takes about 60 seconds