Crawl Cove uses a small, precise vocabulary. Get these dozen terms straight and every other guide reads faster — the words mean the same thing on every page of the app. This is a skim-and-reference page; jump to the term you need.
Core objects
These are the things Crawl Cove stores and works with.
Client : A site — or a set of domains — that you manage. Everything else in the app lives inside a client workspace. Clients are archived, never deleted, so audit history is always preserved. See adding your first client and multiple websites per client.
Audit (crawl run) : One timestamped pass of the crawler over a client's site. It produces the pages and the findings for that run. Audits are versioned, so you can compare one run against another instead of treating each as a throwaway snapshot. See running your first audit.
Finding : A single detected issue, produced by a deterministic check. Each finding has a severity, a list of affected URLs, evidence, and a status — and it is always traceable to the coded rule that produced it, never to an AI guess. See mastering the Findings Explorer.
Check : The coded rule that produces findings — for example, "title too long." Every check has a stable id, a semantic version, a default severity, and tunable threshold parameters. Because checks are versioned, run-to-run comparisons stay honest even after you improve a check. See tuning checks and the audit check reference.
Connection : A per-client integration — a Google Search Console property, a Bing site, and so on — each with its own health status. The Connections page shows you, in one look, whether each integration is working and when it last synced.
Task : A unit of work, usually created from a finding, tracked on a Kanban board. Tasks carry the originating finding's evidence and effort estimate, and pass through a self-verifying Awaiting Verification column. See tasks and verification.
Scoring and prioritisation
How Crawl Cove turns a long list of findings into an ordered to-do list.
Severity : The seriousness of a finding, on a fixed scale: critical / high / medium / low / info. Every check ships with a default severity, which you can override per client.
Impact score : A ranking number computed as severity weight × log1p(GSC impressions on the affected URLs). It sorts "400 findings" into "fix these 9 first," because it weights issues by the real search traffic sitting on the affected pages. Impact requires Google Search Console; without it, findings rank by severity alone. See prioritising with impact score.
Status (of a finding) : Where a finding sits in its life cycle: open / in progress / ignored / fixed / regressed. "Ignored" always carries a recorded reason, so the queue stays honest rather than rotting.
Delta chips : The markers on the Audits page that compare a run to the one before it: ▲ new, ✓ fixed, and ● persisting findings. This is the "prove the fix worked" feature — it turns an audit from a one-off PDF into a versioned history you can put in front of a client. See proving the fix.
How the crawler works
Terms you'll meet when you run and read an audit.
Tier-1 vs Tier-2 : The two passes of the crawler. Tier-1 is a fast static-HTML crawl that is always on. Tier-2 is a sampled, rendered pass (via Playwright) that runs JavaScript so it catches JS-injected content and schema — it's optional and needs a chromium binary installed. See how the crawler behaves.
The principles behind the words
Two phrases recur across the guides because they're the product's positioning, not just jargon.
Deterministic finding : A finding produced entirely by a coded check function — same input, same result, every time. No AI ever invents, removes, or re-scores a finding. The optional LLM layer only explains findings in plain English and drafts report summaries; it is interpretation, never detection. This is why you can stand behind every number you show a client.
Free data sources : Crawl Cove runs on its own crawler plus Google Search Console, Bing Webmaster Tools, Chrome UX Report field data, Lighthouse, and (optionally) Open PageRank — all free. There's no crawl quota and no metered API to top up, so the savings pass straight through to you and your clients. See privacy and your data.
Local-first
: Every byte — the database at <userData>/crawlcove.db and the encrypted secrets
in <userData>/secrets/ — lives on your own machine. Nothing is uploaded to a
vendor backend. See installing Crawl Cove.
Next
- What is Crawl Cove? — the overview these terms describe.
- Running your first audit — see the vocabulary in action.