Published on

June 17, 2026

Updated on

August 13, 2026

Marketing tracking plan: moving from decorative reporting to decision-making 2026

Measure less, decide better: a clean tracking plan avoids unnecessary dashboards.

Request Logiks audit

Marketing data analysis screen, illustration of a decision-making tracking plan.

Good tracking is not about producing more numbers.
It serves to reduce bad decisions.

1. Key figures

Number	What to Understand	Source
2 or 14 months	In standard GA4, user-level data retention is set to 2 or 14 months. Measurement governance must therefore plan for what we keep, export or compare.	Google Analytics Help - Data retention
13 months	The CNIL sets a limited lifespan of 13 months for certain audience measurement trackers falling within the scope of the consent exemption.	CNIL - Audience measurement tools
100 characters	Google Analytics indicates that event parameter values are generally 100 characters maximum, with limits specific to `page_title`, `page_referrer` and `page_location`.	Google Analytics Help - Automatically collected events
5 UTM families	Google documents settings `utm_source`, `utm_medium`, `utm_campaign`, `utm_term` and `utm_content` to identify campaigns that are sending traffic.	Google Analytics Help - URL builders
SHA-256	Google Ads enhanced conversions use a SHA-256 hash on first-party data like email before sending to Google.	Google Ads Help - Enhanced conversions
Data layer	Google describes the data layer as an object used by Tag Manager and gtag.js to pass events and variables to tags.	Google Tag Platform - The data layer
Raw events	The BigQuery GA4 export allows you to query raw events in SQL and combine them with other sources.	Google Analytics Developers - BigQuery export

2. Introduction

Decorative reporting has a formidable elegance. He reassures in meetings, he completes the slides, he gives an impression of control. Sessions, views, clicks, rates, scroll, impressions, conversions, cost per lead: everything seems to be available. However, when a real decision arrives, the system slips away. Should this campaign be cut? Rewrite this landing page? Strengthen SEO? Change the form? Call leads faster? Move the budget? Nobody really knows.
The verdict is simple: data that does not change any decision is an expensive decoration.

We approach this measurement architecture as a proof architecture. We don't start with tags. We start with the decisions: what do we want to arbitrate, how often, with what acceptable margin of error, in which tool and with which owner? Only then come GA4, Google Tag Manager, CRM, UTMs, consent, dashboards and possible export to BigQuery.

A useful device does not promise omniscience. It promises better: fewer signals, better named, better linked to reality.

3. Symptoms: when tracking measures everything except the decision

We recognize weak tracking by its apparent generosity. It measures all clicks, but does not distinguish low interest from commercial intent. It reports forms, but not their CRM quality. It follows campaigns, but the UTMs change depending on the teams. It shows conversions, but no one knows if they correspond to a qualified lead. He stacks dashboards, but none clearly says what to do Monday morning.
Everything is moving. Nothing decides.

Weakness rarely comes from a single tool. It is born from a poorly governed chain: vague naming, events without owner, historical tags, misunderstood consent, absent data layer, unconnected CRM, poorly marked campaigns, dashboards designed to show rather than decide. The problem is not that the company lacks data. It is often the opposite: it lacks a model to eliminate the noise.

As in a full-service kitchen, too many unprioritized tickets do not create control. They create delays.

4. Actors: GA4, GTM, CRM, Ads, CNIL, BigQuery, business teams

Serious action begins with distributing responsibilities. Otherwise, the analyst ends up alone repairing decisions made elsewhere.

Actor	Role in the plan	Question to be decided
Management	Defines priority decisions and piloting cadence.	Which arbitrages should become more reliable?
Marketing	Campaign content, content, forms, landing pages.	What signals trigger action?
Acquisition paid	Tracks budgets, costs, conversions, UTM, offline conversions.	What data feeds media optimization?
SEO	Connects visibility, entry pages, queries, meshing and conversion.	Which organic pages create measurable value?
GA4	Collection of events, parameters, key events and routes.	Which events are really worth setting up?
Google Tag Manager	Deploys tags, variables, triggers, data layer and consent.	Who validates each tag before publication?
CMP / legal	Frames consent, tracers, purposes and proof of choice.	Which tags expect user consent?
CRM	Qualifies the actual business value.	Which leads become opportunities or customers?
BigQuery / data	Preserves and queries raw events as necessary.	When is standard reporting no longer enough?

The measurement gains in robustness when each signal has an owner, a use and a limit. Without this, the dashboard turns into a shadow theater.

5. Definition: what a marketing tracking plan is

A marketing tracking plan is an operational document that describes the measurement objectives, the events to be collected, the associated parameters, the key conversions, the UTM rules, the consent conditions, the tools involved, the responsibilities, the quality controls and the decisions that each data must inform.

This definition matters. It rules out two abuses: the technical file incomprehensible for marketing and the marketing table too vague for developers. A good measurement document circulates the same truth between the business team, the integrator, the data analyst, CRM and management.
This is not a list of tags. It is a decision contract.

6. Why this is becoming a priority in 2026

Google Analytics 4 has installed event logic. Google explains that some events are collected automatically, while recommended events require additional configuration to measure more behaviors and produce more useful reports. This freedom is precious. It backfires on the team if each interaction is given its own name.

The conservation constraint reinforces the need for method. With standard retention adjustable to 2 or 14 months, the business needs to know what it wants to compare over time. For long cycles, CRM arbitrages or advanced analyses, the BigQuery export may become relevant. But an SQL database does not repair a fragile nomenclature.

The consent framework also weighs on architecture. The CNIL strictly regulates trackers, and Google consent mode adapts the behavior of tags according to user choices without replacing the consent banner. Modern tracking must therefore reconcile measurement, compliance, performance and business readability.
Data is no longer a by-product of the site. It forms a commercial infrastructure.

7. SEO, paid, CRM and consent: the same chain of proof

Marketing likes to break up tools: GA4 for audience, Google Ads for campaigns, Search Console for SEO, CRM for sales, Looker Studio for reporting. The client does not go through tools. He is going through an experience.

A visitor can discover a page through Google Search, return through a LinkedIn campaign, download a guide, receive an email sequence, fill out a form, speak to a salesperson, then sign three weeks later. If the tracking plan stops at the first form, the company only measures one fragment. If the CRM ignores the source, it loses the real value. If consent blocks certain tags, it must assume this in the interpretation.

Tracking governance serves to connect these readings without promising perfect attribution. She names the places where the evidence is strong, those where it is partial, those where it is lacking.
No magical truth. A chain of proof.

8. Recommended method: 10 blocks to build useful tracking

The method recommended below is not a proprietary method Logiks. It is based on best practices GA4, Google Tag Manager, consent, UTM campaign, CRM, data layer and marketing management.

We move forward like an architect: use, plan, foundations, networks, control. The decoration comes next.

8.1. Start from decisions, not tools

List the decisions that tracking must improve: cut a campaign, increase a budget, prioritize a page SEO, revise a form, modify an offer, call back a lead more quickly, test a landing page, invest in a channel. Each decision must have a main signal, an alert threshold and an owner.

If no decision is associated with a piece of data, this piece of data deserves to be discussed. Measuring always costs something: time, complexity, consent, performance, maintenance.

The right question is not "can we measure it?"
The right question is "what will we do if this number changes?"

8.2. Define an event taxonomy

GA4 allows you to send automatic, recommended and personalized events. Google-recommended events include useful behaviors like generate_lead, sign_up, purchase, login, search or e-commerce actions. Custom events should remain consistent with your business model.

We therefore define a taxonomy before touching on the tags: event name, description, trigger, parameters, source tool, page concerned, consent required, priority level, owner. This table forms the measurement dictionary.
Without taxonomy, the measure ages like an attic.

8.3. Separate events, key events and KPIs

Not all events deserve to be key events. Google reminds us that a collected event can become a key event when it measures an important action. On a B2B site, a quote request, an appointment or a qualified form can be conversions. A click on an FAQ accordion often remains a secondary signal.

The KPI must go up another level: cost per qualified lead, commercial transformation rate, share of leads from SEO, volume of requests per offer, processing time. The event describes the action. The KPI illuminates the arbitrage.
Confusing the two creates noise.

8.4. Build a strict UTM nomenclature

Google explains that UTM parameters allow you to identify the campaigns that send traffic: source, medium, campaign, term, content. This simplicity quickly turns into chaos if the team writes LinkedIn, linkedin, li, paid-social and paidsocial depending on the day.

A specific convention imposes rules: lower case, separators, authorized source, authorized medium, campaign name, country, audience, format, date if necessary. It also plans who creates the links, where they are stored and how expired campaigns are archived.
A poorly named UTM does not break the site. It spoils the decision.

8.5. Formalize the data layer

Google describes the data layer as the object that transmits information, events and variables to Tag Manager or gtag.js. For an advanced marketing site, it serves as a clean base between the interface, tags and measurement tools.

Useful information is pushed there: page type, offer, sector, persona, form ID, connection status, amount, content category, level of intention, result of an action. Names must remain stable, documented, tested.
The data layer is plumbing. Invisible, but decisive.

8.6. Integrate consent from the plan

Consent should not be added after tracking as a legal veneer. Each tag has a purpose: audience measurement, advertising, personalization, security, functionality. Google consent mode receives the choices transmitted by your banner and adapts the tags; it does not replace the CMP. The CNIL, for its part, regulates the exemption conditions for certain audience measurement tools.

We therefore document for each event or tag: purpose, basis of consent, behavior before choice, behavior after refusal, expected proof, legal or marketing owner.
No abstract conformity. Verifiable behaviors.

8.7. Linking CRM and sales qualification

B2B tracking that stops at form submission rarely measures value. It must link source, campaign, entry page, requested offer, hidden field, CRM status, qualification, opportunity, customer, potential figure or signed figure.

Google Ads provides enhanced conversions and lead-enhanced conversions, with SHA-256 hashing of first-party data when the use case warrants it. The operational principle remains broader: business value must return to measurement, even imperfectly.
The lead is not a final conversion. This is the start of an audit.

8.8. Plan quality controls

The device is not finished when the tags are published. You need a recipe: GTM preview, debug GA4, consent refused/accepted test, forms, UTM, cross-domain, pages 404, duplicate events, key events, CRM data, internal traffic, loading speed.

Each check requires proof: capture, URL tested, date, GTM version, responsible, expected result, anomaly, correction. This rigor seems heavy. It avoids weeks of false reporting.
The data is earned by the recipe.

8.9. Define dashboards by decision

A dashboard is not intended to show everything. It must answer one question: do paid campaigns create qualified leads? Does SEO generate requests on priority offers? Do resource pages feed into the CRM? Do forms break on mobile? Does consent significantly modify reading?

Each table should have an audience, a cadence, three to seven indicators, a source, a threshold and a possible action. The rest is for exploring, not piloting.
The dashboard should not be hypnotizing. It must trigger.

8.10. Provide BigQuery only when the need is real

BigQuery becomes useful for retaining raw events, joining GA4 with CRM, enriching reports, logging beyond standard limits or querying complex paths. However, the tool requires skills, costs, rights and governance.

We recommend considering it when standard reporting is no longer sufficient to address arbitrages. Not to give a data aura to tracking that is still poorly named.
Sophistication comes after clarity.

9. Logiks advice: measure less, decide better

We recommend writing this roadmap as a management document, not as a technical inventory. The technique must be impeccable, but it must serve a question that can be read by a manager: where to invest, what to correct, what proof croire?

First tip: limit key events. Three to eight well-chosen conversions are better than thirty micro-shares. A conversion must involve an intention, a cost or a value.

Second tip: create a shared dictionary. Even simple, it contains name, definition, tool, owner, source, use, consent and status. Without a dictionary, the team will depend on one person's memory.

Third tip: link the forms to the CRM. An unqualified submission can flatter GA4 and tire salespeople. The arbitrage is taken on quality, not just volume.

Fourth tip: do not mix diagnosis and management. The diagnosis can be detailed, technical, exploratory. Piloting must remain short, regular, arbitrage oriented.

Finally, we advise accepting the documented imperfection. Honest tracking with explicit limits is better than a shiny dashboard that hides its blind spots.

10. Decision grid: event, conversion or noise?

Signal	Event	Key event	Recommended treatment
Sending contact form	To keep	Conversion if real commercial demand	Exclude testing and spam
White paper download	To keep	Conversion if strong intention or CRM scoring	Consolidate weak resources
Click CTA	To be continued on strategic pages	Rarely	Group decorative clicks
Scroll 90 %	Content diagnosis	Rarely	Delete if no decision-making use
Phone click	To keep	Conversion for local or commercial activity	Exclude if not used
Account login	To keep	According to model SaaS	Outside scope if not marketing
Purchase or payment	To keep	Priority conversion	No ambiguity
Consent refused	Compliance diagnosis	Not a conversion	Keep to interpret the measurement
Page 404	Site quality	Not a conversion	Keep to detect friction
Opening menu	Generally useless	Not a conversion	Often ignored

This grid is not universal. It forces the arbitrage. Good measure does not keep everything; she knows why she keeps a signal.

11. Common mistakes: nine ways to fabricate useless data

First mistake: creating events without an associated decision. We measure, then we seek justification afterwards.
Second drift: mark too many key events. The word "conversion" then loses its value.

Third weakness: letting the teams create their UTMs each on their own. Reporting turns into a mosaic of variants.

Fourth pitfall: ignoring consent in technical documentation. The measurement discrepancies become incomprehensible.

Fifth risk: forgetting CRM. Marketing then optimizes for form, not revenue.

Sixth confusion: confusing data layer and catch-all. An unnecessary variable is more debt.

Seventh mistake: publishing a GTM container without a recipe. Maybe the tag works. The decision remains uncertain.

Eighth trap: build a single dashboard for everyone. Management, acquisition, SEO and trade do not ask the same questions.

Last point: do not archive old conventions. Without history, the team no longer knows why a name exists.

12. Action Plan 30 / 60 / 90 days

12.1. Within 30 days

list marketing decisions to improve;
inventory existing tags, events, conversions and dashboards;
audit UTM, forms, CRM, consent and GTM;
set 5 to 10 key events maximum;
create the tracking dictionary;
document naming rules;
identify unused data.

We first try to remove the fog. No need to add screens.

12.2. Within 60 days

reconstruct the taxonomy of events;
formalize the data layer;
correct priority UTMs;
link forms and CRM;
test consent and critical tags;
create dashboards by decision;
train teams that publish campaigns.

The measurement is starting to become usable.

12.3. Within 90 days

set up a monthly recipe;
document GTM changes;
bring together leads and commercial quality;
decide if BigQuery is necessary;
archive obsolete tags;
create a marketing + data + commerce ritual;
review the KPIs based on the decisions actually taken.

At this point, the plan is no longer a project file. It becomes a driving discipline.

13. FAQ: marketing tracking plan

13.1. What is the difference between a tracking plan and a dashboard?

The tracking document describes what should be measured, how, why, with what rules and for what decision. The dashboard displays a selection of indicators. Without this base, the dashboard often shows figures without guarantee of quality.

13.2. How many events should be tracked in GA4?

There is no universal number. For an SME marketing site, it's best to start with a few well-defined events: forms, high sales clicks, high-intent downloads, calls, registrations, key pages. Depth comes next.

13.3. Should all important events become key events?

Not necessarily. A key event must represent an important action for the company. Many events are used for diagnosis without meriting conversion status. This separation protects the readability of the reporting.

13.4. Why are UTMs often inconsistent?

Because they are created urgently, by several people, without a common agreement. Source, medium, campaign, content and term must follow shared rules. Otherwise, GA4 receives variants that report frag.

13.5. Is the data layer necessary for a marketing site?

Not always. It becomes useful when events must carry reliable information: page type, offer, form ID, value, status, category, persona, campaign or business context. The more strategic the digital system becomes, the more structuring the data layer becomes.

13.6. Is a tracking plan also a subject RGPD?

Absolutely. Tags, purposes, consents, tracers, durations and transfers must be documented. The CNIL regulates in particular the audience measurement and the conditions of exemption from consent. The plan does not replace legal analysis, but it makes behavior visible.

14. Conclusion: tracking becomes a management asset

Marketing tracking doesn't deserve its complexity if it doesn't change any decisions. It becomes useful when it links campaigns to leads, forms to CRM, pages to SEO, tags to consent, dashboards to arbitrages.

We are not trying to measure everything. We seek to measure what makes action fairer: cutting, strengthening, correcting, prioritizing, investing.

It’s no longer just a tag plan.
Tracking becomes a management asset: name, link, decide.

15. Main sources

Google Analytics Help - Data retention - accessed on June 17 2026 - https://support.google.com/analytics/answer/7667196
Google Analytics Help - Automatically collected events - accessed on June 17 2026 - https://support.google.com/analytics/answer/9234069
Google Analytics Developers - Recommended events - accessed on June 17 2026 - https://developers.google.com/analytics/devguides/collection/ga4/reference/events
Google Analytics Help - About key events - accessed on June 17 2026 - https://support.google.com/analytics/answer/9267568
Google Analytics Help - URL builders: collect campaign data with custom URLs - accessed on June 17 2026 - https://support.google.com/analytics/answer/10917952
Google Analytics Help - Traffic-source dimensions, manual tagging, and auto-tagging - accessed on June 17 2026 - https://support.google.com/analytics/answer/11242870
Google Tag Platform - The data layer - accessed on June 17 2026 - https://developers.google.com/tag-platform/tag-manager/datalayer
Google Tag Platform - Set up consent mode on websites - consulted on June 17 2026 - https://developers.google.com/tag-platform/security/guides/consent
Google Tag Manager Help - Consent support mode - accessed on June 17 2026 - https://support.google.com/tagmanager/answer/10718549
Google Analytics Developers - BigQuery export for Google Analytics - last updated on 20 August 2025, accessed on 17 June 2026 - https://developers.google.com/analytics/bigquery/overview
Google Ads Help - About enhanced conversions - accessed on 17 June 2026 - https://support.google.com/google-ads/answer/9888656
Google Ads Help - About enhanced conversions for leads - accessed on 17 June 2026 - https://support.google.com/google-ads/answer/15713840
CNIL - Cookies: solutions for audience measurement tools - consulted on 17 June 2026 - https://www.cnil.fr/fr/cookies-solutions-pour-les-outils-de-mesure-daudience
CNIL - Sheet n°16: Use analytics on your websites and applications - consulted on 17 June 2026 - https://www.cnil.fr/en/sheet-ndeg16-use-analytics-your-websites-and-applications