A marketing measurement plan for growing businesses: what to count, who owns it, and what it decides

What a measurement plan is, and is not

A measurement plan is the document that sits between your business goals and your analytics tools. It says what you are trying to achieve, how you will know, what exactly gets counted, who is responsible for each number and what you will do differently depending on what the numbers say.

It is not a tracking specification, although it produces one. It is not a dashboard, although a dashboard is built from it. And it is not a list of everything GA4 can collect. That list is long, mostly free and mostly useless to an executive team.

The plan is usually four to six pages. The organizations that need it most are the ones that have been growing for a few years: several channels, a couple of suppliers, a CRM someone set up once, and a leadership team that has stopped trusting the monthly report because three people produce three different lead counts.

This article covers the plan itself: the thinking and the definitions. The configuration that follows (events, key events, consent, cross-domain tracking and the rest) is covered in our GA4 setup and audit checklist. Do them in that order. Configuring first and deciding later is how most broken analytics setups were born.

Start with decisions, not metrics

The fastest way to write a useless plan is to start with the metrics you already have. The useful way is to start with the decisions leadership makes about marketing in a typical year, and work backward.

For most established organizations the list is short:

• How much to spend on marketing next year, and whether to change it mid-year.
• Which channels get more budget, which get less and which get cut.
• Whether the website is doing its job or needs work.
• Whether a specific campaign, program or supplier is worth continuing.
• Whether marketing is producing enough of the right kind of demand for sales, or for operations, to handle.

Write each decision down, then ask: what would we need to see to make this decision with confidence? That answer is your KPI. Everything else is a diagnostic metric, useful when something goes wrong, but not something leadership needs to look at monthly.

A useful test for any proposed metric: if it doubled next month, what would we do differently? If the honest answer is nothing, it is not a KPI.

Define KPIs and the conversion taxonomy

Pick three to six KPIs. More than that and nobody remembers them. Fewer than three and you are probably measuring only volume, not quality or cost.

A sensible set for a service business might be: qualified inquiries per month, cost per qualified inquiry for paid channels, inquiry-to-opportunity rate, and revenue from marketing-sourced opportunities over a rolling period. A tourism operator might use bookings, booking value and share of bookings from outside the province. The right set depends on how the business actually makes money.

The conversion taxonomy

This is the part most organizations skip and later regret. A conversion taxonomy is a written list of every action you count, grouped by how much it matters, with a precise definition for each.

1. Primary conversions. The actions that represent real business: a quote request, a booking, a purchase, a demo request, a job application if recruitment is a goal. These become key events in GA4 and conversions in the ad platforms.
2. Secondary conversions. Actions that show intent but are not business yet: a click on a phone number, a brochure download, a newsletter signup, a visit to the pricing or careers page. Track them, report them separately, and do not let ad platforms optimize toward them unless you have a reason.
3. Engagement signals. Scroll depth, video plays, time on key pages. Diagnostic only. They help explain why primary conversions moved.

For each conversion, write the exact definition. “Form submission” is not a definition. “A successful submission of the quote form on any page, counted once per submission, confirmed by the form platform’s success response, excluding submissions flagged as spam” is a definition. It sounds pedantic until two suppliers report different numbers and you need to know who is right.

Name things once

Agree on event names and stick to them across GA4, the ad platforms and the CRM. If the website calls it generate_lead, the CRM calls it Web Inquiry and the agency report calls it Leads, someone will eventually add them together. A single naming table in the plan prevents that.

Assign owners and a reporting cadence

Every KPI needs one owner. Not a department, a person (by role in the plan, so it survives turnover). The owner is responsible for the number being correct, for explaining movement, and for raising a flag when the measurement breaks.

Ownership usually splits like this:

• Marketing lead owns inquiry volume, channel performance and cost per inquiry.
• Sales lead owns qualification rate, opportunity rate and the accuracy of CRM stages.
• Whoever manages the website owns tracking integrity: that the events still fire after every release.
• Finance or the executive sponsor owns the revenue figure and the definition of marketing-sourced revenue.

Then set the rhythm. A cadence that works for most organizations is a short weekly check for anyone running paid campaigns (is anything broken, is spend pacing correctly), a monthly review of KPIs against targets with a written note on what changed and why, and a quarterly review where budget and channel decisions are actually made. The monthly review should rarely move budget. The quarterly one should almost always consider it.

For seasonal businesses, which describes a large share of Atlantic Canadian tourism, recreation and hospitality, compare against the same period last year rather than last month. A ski hill’s October will always look terrible next to its February, and that tells you nothing.

Map each metric to the decision it enables

This is the section that makes a measurement plan different from a list of metrics. For each KPI, write one line stating the decision it informs and the threshold that triggers action.

An illustrative example. Say a 40-person engineering firm with offices in Fredericton and Halifax runs search ads, LinkedIn and a referral program. Its plan might read:

• Qualified inquiries per month: informs overall budget. If below target for two consecutive months, review channel mix at the next quarterly meeting rather than waiting for year-end.
• Cost per qualified inquiry, by paid channel: informs channel allocation. If one channel runs well above the others for a full quarter with no improvement in deal quality, shift budget.
• Inquiry-to-opportunity rate, by source: informs whether marketing is attracting the right buyers. If a source produces volume but rarely converts to opportunities, fix the targeting or the offer before adding budget.
• Marketing-sourced revenue, rolling twelve months: informs the annual budget conversation. Reported with its caveats, never as a precise return figure.

The thresholds will be wrong at first. That is fine. Writing them down forces the conversation, and you can adjust them after a quarter of real data. What you cannot do is make a defensible decision with no threshold at all.

Connecting inquiries to opportunities and revenue requires the CRM to hold source data, which is its own piece of work. We cover it in how to connect marketing reporting to sales outcomes, and the lead-quality side specifically in how to measure lead quality from paid campaigns.

Data sources, consent and caveats

The plan should state, for each KPI, where the number comes from. GA4 for website behaviour. The ad platforms for spend. The CRM for qualification and revenue. The booking or point-of-sale system for transactions. When two sources disagree (and GA4 and an ad platform always disagree, because they count differently), the plan says which one is authoritative for which purpose.

It should also be honest about what will not be measured precisely. Web analytics in Canada increasingly runs behind a consent banner, and visitors who decline analytics cookies are either not counted or counted only in modelled form, depending on how consent is configured. Federal privacy law expects meaningful consent, and Quebec’s Law 25 sets stricter expectations for anyone serving Quebec residents, which includes many New Brunswick organizations with bilingual audiences. Your legal counsel should interpret what applies to you. The measurement plan should simply record how consent works on your site and what that means for the numbers, so nobody is surprised.

Write the caveats down in plain language. Phone calls that start on the website but are not tracked. Referrals that show up as direct traffic. Walk-in customers who found you on Google Maps. A plan that admits its blind spots is more credible to a board than one that pretends to see everything.

A one-page template you can adapt

The full plan may run several pages, but it should open with a single page leadership can read in two minutes. That page holds:

1. Business goals for the year, in one or two sentences each.
2. KPIs, three to six, each with its target, owner and source of truth.
3. Primary conversions, each with its exact definition.
4. Decision rules: the threshold for each KPI and the action it triggers.
5. Cadence: weekly, monthly and quarterly, and who attends each.
6. Known gaps: what is not measured and why.

Behind that page sit the detail: the full conversion taxonomy, the naming table, the UTM conventions for campaign links, the consent configuration and the tracking specification your developer or agency will build from. The executive dashboard is then just this one page, rendered with live numbers.

Tourism and public bodies can use exactly the same structure; the KPIs change, the discipline does not. Our piece on what a tourism board should actually measure walks through that version.

Common mistakes

• Too many KPIs. Twenty metrics on a monthly report means none of them get discussed.
• Every action counted as a conversion. If newsletter signups and quote requests are added together, the total is meaningless and the ad platforms will chase the cheaper one.
• No owner for tracking integrity. A website release quietly breaks the form event, and nobody notices for two months.
• Definitions that live in someone’s head. When that person leaves, the numbers change meaning without anyone saying so.
• Precision theatre. Reporting a return on marketing to two decimal places, built on attribution that cannot support it.

In our analytics and measurement work, the plan is always the first deliverable, agreed in writing by marketing, sales and leadership before anything is configured. It is the least glamorous document in the engagement and the one clients refer back to most.