An agency dashboard we once inherited tracked 43 SEO metrics. Week over week. With sparklines. The client churned anyway, because none of the 43 was "revenue from organic" — and the three numbers that predicted the churn (declining clicks on money pages, flat referring domains, a keyword portfolio stuck on page two) were buried under forty other charts nobody read.
Most SEO measurement fails in that exact way: not too little data, too much of the wrong kind. Here's the short list that's actually worth your attention, sorted by what each number can and cannot tell you.
Three layers of metrics
Every useful SEO metric belongs to one of three layers:
- Outcome metrics — did the work pay off? Organic traffic, organic conversions.
- Predictive metrics — will it pay off soon? Authority, ranking distribution, share of voice.
- Diagnostic metrics — why is it (not) working? Engagement, index coverage, CTR.
The layers answer different questions on different timescales. Mixing them — treating a diagnostic like an outcome, or optimizing a predictive metric as if it were the goal — is where most bad SEO strategy comes from.
The outcome layer
1. Organic traffic (yours: measured; competitors': estimated)
For your own sites, Search Console clicks are the truth — use them. For any site you don't own, all traffic figures are third-party estimates, and you should compare estimates only against other estimates from the same source. Watch the trend and the mix: total visits alongside organic visits tells you whether growth is search-driven or something else entirely.
2. Organic conversions
The only metric leadership actually cares about. If you can't attribute conversions, count proxies with intent: pricing-page sessions from organic, signups with organic first-touch. Traffic that converts nothing is a vanity graph.
The predictive layer
3. Domain Rating (used correctly)
DR is a 0–100 comparison of link-profile strength. It predicts competitive capacity: a DR 25 site pitching a keyword whose top ten are DR 70+ is bringing a knife to an artillery duel. What DR is not: a KPI to farm. Chasing DR through junk links moves the score without moving anything the score exists to predict — and link-spam signatures age badly. Use it to pick fights you can win, and to qualify link prospects.
4. Referring domains, trended
The input behind DR, and the better week-to-week signal because it moves first. Count unique linking domains, not raw links — the thousandth link from one site is worth almost nothing. Flat referring domains with growing content output usually means the content isn't reference-worthy.
5. Ranking distribution
Not "we rank #1 for X" — the histogram: how many tracked keywords sit at 1–3, 4–10, 11–20. Movement from page two to page one is the strongest leading indicator of a traffic jump, and it shows up weeks before the traffic does.
The diagnostic layer
6. Bounce rate & engagement — with intent context
A 90% bounce on a dictionary-style page is a success: the visitor got the answer. A 90% bounce on a pricing page is a fire. Bounce rate, pages per visit and time on site mean nothing as absolutes; they mean a lot when a page's number diverges from its intent class or its history.
7. Index coverage
Pages that aren't indexed don't rank — obvious, routinely unchecked. A periodic index check on your money pages catches deindexing accidents (noindex tags shipped to production, canonical mishaps) days before the traffic chart does.
8. CTR on ranking keywords
Ranking #4 with a title nobody clicks is a rewrite opportunity that costs an hour. Search Console's query report, sorted by impressions × low CTR, is the cheapest wins list in SEO.
Stop tracking these
- Keyword density — died with dial-up. Anyone selling density audits is selling audits.
- Total keywords ranked — trivially inflatable with zero-value long tail. Weight by volume and intent or ignore it.
- #1 positions on vanity terms — a #1 on a term with no volume, or on your own brand name, decorates a report and changes nothing.
- Raw backlink counts — see referring domains, above. One number is a strategy signal, the other is noise.
The five-number monthly dashboard
| # | Metric | Layer | Question it answers |
|---|---|---|---|
| 1 | Organic conversions | Outcome | Is SEO making money? |
| 2 | Organic traffic trend (both estimates for competitors) | Outcome | Is visibility compounding? |
| 3 | Referring domains, new this month | Predictive | Is authority still being earned? |
| 4 | Keywords at positions 4–15 | Predictive | What's about to break onto page one? |
| 5 | Index coverage on money pages | Diagnostic | Is anything silently broken? |
Five numbers, one screen, and every one of them either predicts, explains or is the result. For the competitor half of that dashboard — DR, referring-domain proxies, and both traffic estimates in one call — that's the exact field set our free checker and API return. For benchmarking those numbers against your industry, we keep typical ranges here.