Analytics that records real outcomes.

• Choose analytics implementation when the numbers cannot be trusted.

  If reports disagree with CRM or order records, or no key events exist, measurement is fixed before any conversion work, because every later decision reads from it.

• Choose conversion rate optimization when measurement already holds up.

  When key events are reliable and the question has become why visitors do not complete them, the work moves to research, hypotheses and testing.

• Choose technical SEO when pages are missing from Google.

  Search Console reports indexing problems as well as performance. When pages are not indexed, crawling and rendering come first, and measuring their search traffic follows.

A measurement plan lists the business questions analytics must answer, the key events that count as outcomes, the supporting events and parameters that explain them, and the rules for naming and consent. Written before implementation, it is the reference every tag, report and later change is checked against, so the setup does not drift as people change.

A standard Google Analytics property allows up to 30 events to be marked as key events. The limit is rarely the constraint: a handful of key events everyone understands produces reports people use, while a long list turns every report into noise.

• The decisions the business wants to make from the data, in plain language.
• Key events: the few actions that represent value, such as a confirmed inquiry or a purchase.
• Supporting events that explain the path, such as starting a form or viewing a product.
• The parameters each event carries, and which need custom dimensions to appear in reports.
• The naming convention, and the recommended Google events used in place of custom names.
• How each tag behaves under each consent choice, as the business has decided.
• Data that must never be collected, starting with personal information.
• Who owns the plan and approves changes to it.

Analytics events should be named with one consistent convention, using the recommended event Google Analytics already defines wherever one fits and a custom name only when none does. Google Analytics treats event names as case sensitive, requires them to start with a letter, allows only letters, numbers and underscores, and limits names to 40 characters.

Case is the easiest way to split data by accident: Google Analytics treats my_event and My_Event as two distinct events. A convention written into the measurement plan, such as lowercase words separated by underscores, keeps one action under one name.

Recommended events have predefined names and parameters, and are designed to unlock existing and future reporting capabilities. A submitted form or request for information maps to generate_lead, a completed purchase to purchase, and an item added to a cart to add_to_cart.

Consent mode lets Google tags adjust their behavior to the choice a visitor makes. It has two implementations: in basic consent mode, tags do not load until consent is granted and nothing is sent before then; in advanced consent mode, tags load when the page opens and, while consent is denied, send measurements without cookies.

The consent types in consent mode separate analytics from advertising. analytics_storage covers storage such as cookies related to analytics, while ad_storage, ad_user_data and ad_personalization cover advertising storage, sending user data for advertising, and personalized advertising. A visitor can accept one and decline the others, and each tag follows the choice that applies to it.

Google Analytics can model the behavior of visitors who decline analytics cookies, but only when a property meets minimum thresholds, including at least 1,000 events per day with analytics storage denied for at least seven days, and meeting them does not guarantee eligibility.

Which visitors must be asked, and in which regions, is a legal question the business answers with its adviser. Consent mode is the technical implementation of that decision. A banner on its own does not make a website compliant, and no compliance outcome is promised.

Analytics data can be trusted once each tag has been tested before publishing, the events arriving in the property have been checked against the plan, and the totals have been reconciled with a source the business already relies on, such as CRM or order records. Differences that remain are explained and documented rather than ignored.

Some differences are expected, and documented rather than fixed. Search Console clicks and Google Analytics sessions are calculated differently, visitors opting out of tracking can skew Google Analytics data, and Google Analytics tracks traffic only from users who have JavaScript enabled.

• Tags tested in Tag Manager preview mode before a container is published.
• No event fired twice, for example by a hard-coded Google tag and a Tag Manager tag together.
• Internal traffic defined, so the team does not count its own visits.
• Payment providers and other unwanted referrals excluded, so they do not replace the real source of a visit.
• Cross-domain measurement set up where a journey crosses root domains, which otherwise counts one visitor as two users.
• No personal data in page URLs, page titles, event parameters or site search terms.
• Purchases carrying a transaction ID, which helps avoid duplicate purchase events.
• Key event totals compared with CRM or order records for the same period.

Search Console shows what happened in Google Search before a visit: the queries a page appeared for, impressions, clicks, click-through rate and average position. Google Analytics shows what visitors did after they arrived. Linking the two adds Google Organic Search Queries and Google Organic Search Traffic reports to a Google Analytics property.

Search Console counts a click each time a user clicks through to the site from Google Search results, an impression each time the site appears in results, click-through rate as clicks divided by impressions, and average position as the average position of the topmost result from the site. Clicks and impressions from AI features such as AI Overviews and AI Mode are included in that search traffic, reported within the Web search type.

The two sources never match exactly. Search Console reports only on the canonical URL Google selected, while Google Analytics reports on any URL carrying the tag, and some queries are omitted from Search Console reports to protect user privacy. Search Console keeps data for the last 16 months, so a baseline is exported before older data is lost.

• Google Analytics setup
• Google Tag Manager
• Search Console
• Measurement plan and event taxonomy
• Data quality and validation

A measurement setup is built in layers, and each one relies on the layers above it. When a report cannot be trusted, the cause often sits in a layer above the report itself.

• Reports that answer business questions

  Key events mark the actions that matter, so reports show which pages, campaigns and channels lead to inquiries or sales instead of stopping at page views.

• One name for one action

  A written convention, and a recommended Google event wherever one exists, keep event names consistent across the site, the container and every report built on them.

• Numbers the team can defend

  Validation and reconciliation against records the business already trusts show where analytics agrees, where it differs and why.

• Less risk in the data

  Personal data is excluded by design, and consent settings decide what each tag may do, rather than a banner that changes nothing.

• A lead generation site with no key events

  Form submissions and booking requests become defined events, and a confirmed inquiry becomes the key event that reports are built around.

• A store with revenue nobody trusts

  E-commerce events are implemented with a transaction ID and a currency, so purchases are not duplicated and revenue can be reconciled with order records.

• A website rebuild

  The measurement plan is written alongside the new templates, so events exist from launch day instead of being reconstructed after a gap in the data.

• Recommended events are used wherever Google Analytics defines one, and custom events only where no other event fits.
• Event names follow the Google Analytics rules: case sensitive, starting with a letter, using only letters, numbers and underscores, and no longer than 40 characters.
• Tags are deployed through Google Tag Manager and tested in preview mode, which shows which tags fired and in which order, before a container is published.
• Non-essential tags fire after the page has loaded and unused tags are removed, because tags and tag managers add work to the main thread.
• Where a site sends a Content-Security-Policy, only the domains the Google tags in use require are allowed, rather than opening the policy wider than needed.
• Consent mode is technical implementation of a decision the business makes with its legal adviser, never legal advice or a compliance guarantee.