How the grader actually scores you.
Every weight, threshold, and dollar coefficient behind the Zayos grader is on this page. If your listing scored a 42, you can read the rule that produced the deduction. The same JSON powers our scoring engine, this page, and the public API endpoint at /api/zayos/methodology.
What we score
The grader runs across five pillars worth a combined 100 points. A listing starts at 100 and loses points for each finding our scan produces. Pillars are weighted by how directly they convert search traffic to paid tickets, not by how much we wish they mattered.
Pillars and weights
| Pillar | Max points | What it covers |
|---|---|---|
Search & discoverability search_results | 25 | Whether a customer searching for your category nearby actually finds you, and whether the listing they see is rich enough to click. |
Reputation & guest experience guest_experience | 25 | What guests say about you publicly — rating, review volume, recency, and the photos that frame the menu. |
Local listings completeness local_listings | 20 | Hours, address, phone, categories, and other facts that need to be present and consistent before Google ranks you. |
Conversion & ordering surface online_ordering | 15 | Once a customer wants to order, how friction-free is the path from your listing to a paid ticket — direct or via marketplace. |
Website fundamentals website | 15 | Mobile viewport, structured data, page weight, and the basics that determine whether your site shows up in search at all. |
How findings deduct points
Each finding the scan produces has a severity. Severity controls how many points come off the total and how many come off the pillar score. Tuned so a typical poor listing (4 high + 2 medium findings) lands in the 25 to 35 score band.
| Severity | Total deduction | Pillar deduction |
|---|---|---|
| high | −14 pts | −9 pts |
| medium | −8 pts | −5 pts |
| low | −3 pts | −2 pts |
Dollar leak math
Every finding has a $/mo range, not a single number. The range is anchored on a baseline of 1,800 online orders per month at a $35 average ticket (~$63,000/mo in revenue). That's where the typical independent restaurant in our South Florida data set sits.
Per finding, the modeled monthly leak is the severity coefficient applied to baseline revenue:
| Severity | Low / mid / high | Modeled $/mo at baseline |
|---|---|---|
| high | 8.0% / 12.0% / 18.0% | $5,040 / $7,560 / $11,340 |
| medium | 3.0% / 5.0% / 8.0% | $1,890 / $3,150 / $5,040 |
| low | 0.5% / 1.5% / 3.0% | $315 / $945 / $1,890 |
Total leak across all findings is capped at 55% of baseline revenue. Catches model-output runaways where eight high-severity findings would otherwise sum to more leak than your business plausibly produces.
Grade bands
- Strong — 80 and up. Small finishing moves only.
- Fair — 60 to 79. One or two pillars are weak; the rest are healthy.
- Poor — 40 to 59. Material leak across two or more pillars.
- Critical — below 40. Fundamentals are missing across most pillars.
What we do not score
- Who built your website. We have no platform-detection logic that penalizes a competing vendor. The grader scores measurable signals only.
- Grid SERP rank. We don't run a grid rank tracker. Search visibility is scored from listing completeness + rating + review-volume signals, not from a synthetic grid.
- NAP consistency across third-party directories. We don't claim to scan Yelp, TripAdvisor, Bing, or every minor directory. Listed and dropped from the rubric rather than zero-weighted.
- Owner response rate. Google Places does not expose owner responses cleanly. Dropped from the rubric.
Machine-readable
The same rubric served as JSON at /api/zayos/methodology. Public, cache-friendly, CORS-open so AI search engines can cite the methodology directly. Bump in lockstep with this page; version stamp is on both.