Measure search recovery after website restoration with three separate records: whether important URLs can be crawled and indexed, whether relevant searches produce visibility and visits, and whether those visits lead to genuine enquiries. A restored page, a submitted sitemap or a single search result does not prove that rankings or customer demand have recovered.
The baseline matters. Compare the restored site with what actually happened before the loss, not an invented traffic target. If historical analytics are unavailable, start a dated post-restoration baseline and state that limitation. Do not manufacture a before-and-after improvement from two unrelated periods.
Write down what changed and when
Maintain a short chronology: the last known working date, the outage or loss period, the date essential pages returned, and later content or technical changes. Separate the site’s recovery from advertising campaigns, seasonality, tracking changes and platform migrations.
Choose the pages that matter commercially: the homepage, important services, useful guides and legitimate conversion routes. Retain the original URL list. If approved address changes occurred, map old and new destinations before comparing page-level reports. Otherwise, a move can look like a disappearance from the old URL and unrelated growth at the new one.
Define what counts as a lead before looking for success. A form view is not an enquiry. An enquiry is not an order. Record verified, relevant enquiries and paid projects separately, using only information the business is entitled to process.
First checkpoint: technical availability
Inspect important pages directly and with the search engine’s tools. Confirm the intended HTTP response, public content, canonical, robots directives and internal links. A URL can return a successful response while displaying a server error message, an empty template or a different page.
Google’s URL Inspection documentation distinguishes indexed information from the live test. Record which view you checked and its date. A passing live test does not guarantee indexing. An indexing request is a request, not proof that the page is indexed or ranks well.
Keep crawlability, indexability and actual indexing as different statuses. Where the evidence is inconclusive, record it as unresolved rather than converting a technical check into an SEO score. This checkpoint should also catch unintentionally exposed staging pages and old preview destinations.
Second checkpoint: relevant search visibility
Use Search Console’s page, query, country and device filters consistently across comparison periods. Separate brand searches from non-brand restoration searches. A site-name query can recover while relevant service discovery remains weak.
The Search results performance report provides clicks, impressions, CTR and average position. Average position is an aggregate, not a permanent ranking for a keyword. The newest data can be preliminary. Preserve the report filters and export date so another review can reproduce the comparison.
Look at combinations rather than celebrating one rising total. More impressions with fewer clicks may reflect different queries, lower positions or less relevant exposure. A smaller number of visitors can still be valuable if they reach the correct service and submit suitable enquiries. Small samples need cautious interpretation.
Use equivalent-length periods and consider day-of-week and seasonal differences. For a long outage, compare multiple historical windows if reliable data exist. Annotate unusual events rather than quietly removing them. The result should explain uncertainty, not pretend that every later change caused the movement in a chart.
Third checkpoint: enquiries and project outcomes
Check the route from an organic landing page to a legitimate enquiry. Confirm tracking permission, event definitions and whether the form or contact method actually works. A success animation does not establish delivery. If delivery is owner-managed or not verified, report that dependency openly.
Use an aggregate lead register with fields for date range, landing-page group, known acquisition source, relevant enquiry count and confirmed project count. Avoid copying customer messages or personal information into a public recovery report. Unknown attribution should remain unknown.
Search Console describes search exposure; analytics describes measured site activity; the business’s enquiry record describes received contacts. Those systems can disagree because their definitions, attribution and consent coverage differ. Reconcile the definitions before diagnosing a technical fault from mismatched totals.
Measure AI visibility without inventing a universal ranking
Google’s June 2026 announcement describes a dedicated generative-AI performance view for impressions, pages, countries, devices and dates. The underlying AI visibility also remains included in overall performance reporting. Check which reporting is actually available to the verified property; do not claim a report exists in an account you have not inspected.
For other answer engines, keep reproducible observations: platform, prompt, date, region or account context when known, response and cited destination. Separate a brand mention from a clickable citation and a citation from an attributable referral. A single answer cannot establish visibility across all users or all relevant questions.
Repeat a small set of relevant prompts at defined intervals instead of changing the questions until a flattering answer appears. Manual samples are observations, not market-share estimates. There is no honest universal percentage that combines every search engine and AI platform into one recovery score.
A compact recovery review record
- Period and context: dates, outage/relaunch events and known tracking limitations.
- URL evidence: inspected page, response, selected canonical and indexed/live-test distinction.
- Search evidence: saved filters, relevant query/page groups and comparable periods.
- Business evidence: received relevant enquiries and confirmed projects, without personal data.
- AI observations: report or reproducible sample, cited page and its specific limitations.
- Next decision: one evidence-supported action, the responsible owner and a review date.
This format prevents a crawl repair from being reported as a ranking win and a ranking observation from being reported as an order. Keep mobile usability and performance visible as independent priorities, even when new content is being published.
The restoration guide’s URL-preservation checks help establish the technical baseline. For a recovery project, Wayback Restore can review surviving sources and rebuilding requirements. Neither restoration nor publishing more articles guarantees a particular search position or enquiry volume.
Questions about measuring recovery
How soon should traffic return?
There is no reliable universal deadline. Review verified technical status, comparable search evidence and business outcomes over time, accounting for the length of the outage and other changes.
Does an AI citation prove the site has recovered?
No. It is evidence of that response in that context. Record the destination and date, then assess repeat observations and attributable referrals without generalizing to every user.
