Log queries and report integrity
Separate Worker subrequests from visitor-facing requests
Reconcile Cloudflare Worker and zone analytics by naming the request layer before adding counts together.
THE ANALYTICS FIELD LIBRARY
Find the question you need to investigate. Each guide connects a dated source, an original explanatory diagram and a concrete acceptance check.
These guides use bounded observations and documented methods. They do not claim that a crawler visit proves an AI recommendation.
Log queries and report integrity
Reconcile Cloudflare Worker and zone analytics by naming the request layer before adding counts together.
Log queries and report integrity
Calculate represented counts and weighted averages using each row’s sample interval instead of applying one assumed sampling factor.
Crawler access and indexing
Trace the enforcing rule when Cloudflare AI Crawl Control settings and observed crawler responses appear inconsistent.
Collection diagnostics
Compare language and representation selection before labeling different responses at one URL as an indexing or analytics defect.
Measurement and interpretation
Trace route navigation to tracker calls when single-page app pageviews appear twice.
Log queries and report integrity
Preserve query scope and uncertainty before treating small changes in Cloudflare GraphQL results as a real traffic trend.
Crawl access and indexing
Keep the uploaded robots file, public response and Google’s cached observation on separate timestamps.
Crawl access and indexing
Test user-agent group selection separately from URL-path matching when robots behavior seems inconsistent.
Crawler access and indexing
Use a dated rule, requested path and response together when investigating crawler requests to a disallowed URL.
Collection diagnostics
Use Range and Content-Range evidence to distinguish partial transfers from a completed download or a reader.
Measurement and interpretation
Compare the emitted event name with the configured goal, including spaces and capitalization.
Crawl access and indexing
Separate crawl access from noindex processing before repairing a public page or removing a URL from search.
Collection diagnostics
Treat Network Error Logging as an additional browser signal and preserve its support and reporting limits.
Measurement and interpretation
Prioritize browser issues using both repeated exceptions and the number of visits affected, with collection limits visible.
Log queries and report integrity
Define a stable ordering and test boundary ties before calling a multi-page GraphQL export complete.
Log queries and report integrity
Prevent a failed Cloudflare analytics query from becoming a zero-traffic chart or a silently incomplete report.
Log queries and report integrity
Separate zone selection from dataset conditions and verify AND/OR logic with a small inclusion matrix.
Log queries and report integrity
Use the Cloudflare Settings node to distinguish unavailable history, query-window limits and page-size limits.
Crawl access and indexing
Do not turn a missing origin-log entry into a successful request; inspect the layers before HTTP.
Crawl access and indexing
Separate common crawlers, special-case crawlers and user-triggered fetchers before interpreting robots behavior.
Crawl access and indexing
Investigate overload signals and recovery without treating a single successful request as restored crawling.
Event validation and attribution
Separate event acceptance from the evidence needed to join a server event to an earlier online interaction.
Event validation and attribution
Use the Measurement Protocol validation endpoint, then verify the actual implementation and reporting destination separately.
Event validation and attribution
Check the documented event and parameter limits before submitting queued Measurement Protocol payloads.
Measurement and interpretation
Check step order and each interval before diagnosing an unexpected funnel drop-off in Umami.
Event validation and attribution
Preserve when an event actually happened and understand the documented handling of backdated Measurement Protocol events.
Collection diagnostics
Add browser request context to an investigation while keeping it separate from visitor identity and verified crawler classification.
Measurement and interpretation
Hold conversion definition and date range fixed before interpreting different channel credit.
Log queries and report integrity
Locate the response layer behind an error trend, including cases where a Worker changes the response returned to the client.
Crawl access and indexing
Inventory filter combinations before changing crawl controls, and test invalid combinations at their own URLs.
Collection diagnostics
Use the browser policy violation and the delivered page policy to locate a tracker execution failure without weakening site security.
Measurement and interpretation
Use Umami identity links carefully and keep a combined visitor timeline separate from session totals.
Collection diagnostics
Name the byte field and compression layer before attributing a bandwidth change to crawler activity.
Collection diagnostics
Separate OPTIONS checks, event requests and accepted records when a browser collector appears to receive twice the expected traffic.
Collection diagnostics
Diagnose a tracker URL that serves an HTML fallback or the wrong media type before treating the response as successful JavaScript delivery.
Collection diagnostics
Distinguish no-cache, no-store and private before changing the caching policy of an authenticated analytics page.
Collection diagnostics
Inspect the method and body at every redirect hop when an analytics endpoint moves or a proxy normalizes its URL.
Measurement and interpretation
Review an Umami Share URL anonymously and distinguish public presentation from authenticated reports.
Collection diagnostics
Keep conditional requests, transferred content and crawler access separate when reporting 304 responses.
Collection diagnostics
Separate proxy-cache residence from the observation window shown inside a saved report.
Product outcomes
A click indicates intent. A completed audit or saved application requires a successful server action.
Evidence integrity
Publication time, source cutoff and observation date describe different events. Read the cutoff before making a current claim.
Access diagnostics
A robots.txt retrieval failure can affect crawling differently from one failed page. Diagnose them separately.
Collection coverage
Newly published URLs are not automatically part of a collector. Reconcile intended pages with configured scope.
Privacy and scope
Public examples need useful locations without query tokens, private report URLs, form contents or visitor identifiers.
Crawler policy
Robots permission and firewall permission are separate. Diagnose the agent and address before widening access.
Reporting methods
A chart may suggest growth because one date covers more hours. Check observation boundaries before comparing daily crawler counts.
Collection coverage
An origin log can miss requests served entirely by a cache. Identify the observation layer before claiming whole-site coverage.
Crawler policy
Separate search, model-development collection and user-directed retrieval before changing OpenAI access rules.
Attribution
A visit without a recorded referrer is unattributed here. Timing alone cannot assign it to an AI assistant.
Observed access cases
A dated own-site case of gateway response failed, with affected public paths, retained request counts and a bounded investigation handoff.
Crawler identity
A user-agent label is a client claim. Keep it separate from verification before reporting that an AI company visited.
Access diagnostics
A larger request count can reflect repeated fetches or operational checks. Investigate the burst before claiming audience growth.
Reporting methods
A percentage needs numerator and denominator from the same population, interval and counting rules.
Collection reliability
An empty chart cannot establish that no crawler visited. Check reception and processing before interpreting zero.
Crawler policy
Different Claude crawler names have different uses. Map the intended access policy to the correct agent.
Measurement operations
Browser pageviews and log requests have different scopes. Reconcile collection rules before treating the difference as missing traffic.
Reporting methods
Your checks can appear in analytics. Label controlled tests before interpreting a new installation as customer demand.
Access diagnostics
A healthy page now and an earlier failure can both be true. Retain the incident and add dated recovery evidence.
Evidence integrity
A chart should be reproducible from its data. Check unit, period and edition before trusting a visual summary.
Product outcomes
A retry should improve event delivery without creating another business outcome. Keep identity separate from delivery attempts.
Access diagnostics
A traffic report shows gateway errors. Locate the failing path and time window before changing bot rules or rewriting content.
Practical delivery
A report should tell the next operator what to inspect and how to verify the result without inventing a root cause.
Use the GEO audit tool for website evidence, or explore the managed AI crawler analytics pilot for an agreed collection and investigation scope.