‹ Blog

We Ran Our Own Tool on Ourselves. The Score Was 57.6.

Dogfooding means running your own tool on yourself and publishing the result — flattering or not. Ours came back 57.6 out of 100. Here's the honest layer-by-layer breakdown.

It's easy to publish a glossy case study with a number that goes up and to the right. It's more useful — and more honest — to run our own tool on ourselves and show you the real result, including the parts that aren't great. So here it is: RankingLocal.ai's own REAL Method score is 57.6 out of 100. That's a "mixed" score, not a victory lap. Here's how it breaks down, and why a young product scores exactly where you'd expect.

The four layers, no spin

The AI Recommendation Rate is four layers, each scored 0–25. Ours:

Add them up: 17 + 11 + 16 + 13.6 = 57.6.

Why this is the right score for us

A 57.6 is not a failure; it's an accurate portrait of a real, early-stage business. And it illustrates something we say constantly: the layers move on different clocks. We could max out Recognize quickly because it's mechanical. We can't fast-forward Evidence, because evidence is earned over time across sources we don't control. A brand-new business with perfect schema and thin reviews scoring in the high 50s is the system working correctly — it's measuring reality, not rewarding effort.

If our tool had handed us a 95, that would be the real red flag. It would mean the score was vanity, not signal.

Note

The most useful number a tool can give you is the unflattering-but-true one. A score you'd be embarrassed to show is worth more than a score designed to make you feel good.

What we're doing about each layer

Publishing the number is only worth something if it drives action. Our own plan maps to the 90-day sequence we'd give anyone:

Why we're telling you our weak number

Two reasons. First, trust: a tool that scores its own maker should be willing to show that score, warts and all. If we hid behind a perfect number, why would you believe the one it gives you? Second, demonstration: 57.6 is probably a more relatable starting point than a polished 90. Most businesses running the free check for the first time land somewhere in the 30s to 60s. Seeing that we're in the same range — and treating it as a to-do list, not a grade — is the entire mindset we want you to take.

The score isn't a judgment. It's a starting line with the lanes marked. Ours says: recognition's handled, content's coming along, fix the crawl issues, and earn evidence patiently. That's a plan, not a verdict.

See your own number

Run your domain through the free AI Visibility check and you'll get the same four-layer breakdown we just showed you for ourselves — no signup. Whatever it says, read it the way we read our 57.6: not as a grade, but as the shortest path to being recommended. If your number surprises you, hello@rankinglocal.ai reaches me directly — I'm genuinely curious what the tool tells people.

Frequently asked questions

What is a good AI Recommendation Rate score?

Anything above about 70 on a layer is strong; 12–19 is mixed with clear fixes available; under 12 means you're effectively invisible on that layer. Most businesses running the check for the first time land somewhere in the 30s to 60s overall. For context, RankingLocal's own REAL score is 57.6 out of 100 — a realistic 'mixed' score for a young business, strong on Recognize (17/25) and weak on Evidence (11/25) because evidence takes the longest to build.

Why would a company publish its own mediocre score?

Trust and realism. A tool that scores its own maker should be willing to show that score honestly — if we hid behind a perfect 95, there'd be no reason to believe the score it gives you. And 57.6 is a more relatable starting point than a polished number, because it's roughly where most businesses begin. The right way to read any score, ours included, is as a prioritized to-do list, not a grade. See your own at /tools/ai-visibility/.

Related reading