Commission useful web analytics
Analysis should inform site use and decisions. Setup, checking and interpretation differ. Dashboards do not prove complete data or correct conclusions. Agree purposes, pages, metrics and limitations.
Define decisions
Do you need sources, content use or ordering journeys? Name the next decision. Many charts do not establish usefulness.
Define visits, actions and outcomes. Contact clicks and valid enquiries differ. Tools may count differently under identical labels.
| Question | Useful result |
|---|---|
| Visit sources | Clear sources and limitations |
| Journey use | Defined steps and analysis |
| Correct measurement | Cases, findings and corrections |
| Next actions | Justified recommendations |
Define sites and data
Record sites, shops, subdomains and journeys. Several sources may describe one visit. Assess measurement and gaps.
Read access may suffice initially. Define implementation for new measurement. Missing historical data is not automatically recoverable.
Assess errors and differences
Check duplicates, gaps, redirects, internal tests and definitions. Browsers, consent and filters affect counts. Server requests are not human visits.
Differences may reflect methods, periods, time zones, attribution or errors. Compare equivalent questions and settings. No general acceptable discrepancy replaces review.
Limit collection to purpose
Check necessary data and risks in URLs, search or parameters. No customer files, passwords or full recordings in the enquiry.
Permissions, notice, consent or prerequisites, retention and sharing should suit the process. Tools, own installation or no cookies do not guarantee legality.
Report limitations
Samples, estimated unique people or suppressed small groups affect presentation. Distinguish these from missing setup. Define metrics, periods and quality.
Attribution does not automatically establish causal impact. Measurement, demand or content changes affect trends. Reports offer observations and useful checks, not unjustified certainty.
Separate technical and interpretation work
Compare sites, events, reports, sources and care. Standards and specific journeys differ. Reports may need preparation and business definitions.
Separate assessment, setup, correction, reporting and interpretation. Licences, storage, processing and queries may add costs. Reuse data and bounded evaluations.
| Item | Question |
|---|---|
| Assessment | Which sites and cases? |
| Measurement | Which events and implementation? |
| Reports | Which questions and definitions? |
| Care | Which analysis and discussion? |
| System | Which licence, processing and limits? |
Define initial work
Describe task, sites, context, questions, tools and event or report count. Zero visits for new sites is possible. Put references and systems in one brief.
Agree plan, checking, changes and acceptance. Record sources, metrics, limits and handover. No completeness or revenue guarantee.
Frequently asked questions
Do I need a new tool?
Not necessarily. Review measurement and questions before licences or work.
Why do visits differ?
Methods, definitions, filters, consent, periods or errors may differ.
Are server requests people?
No. Resources, bots and repetition may be included.
Can missing history be recreated?
Not automatically. Define data and gaps honestly.
Does a tool guarantee compliance?
No. Assess setup, purpose, data and rules.
What should a quote include?
Assessment, plan, events, reports, implementation, interpretation, care and system costs.
