What Are You Trying to Measure?

Your reports have gaps. That’s not the problem.

The problem is that nobody has written down what the gaps are.

This chapter fixes that. By the end of this page you’ll have a list of the questions your reports can’t answer today, and that list is the first draft of your taxonomy. No parameter values yet. No naming conventions. Just the questions.

The Questions Are Always the Same

Every marketing team runs campaigns: ads, emails, social posts, events, partner promotions. And sooner or later, every team gets asked the same questions about them.

Which channel drives the most valuable traffic? Social, email, paid search, or something else entirely?
Which source inside that channel is doing the work? “Social media” is not an answer. LinkedIn versus Facebook versus Twitter is an answer.
Which campaign is converting: the specific promotion, the product launch, the seasonal push?
Some teams need a level deeper: product line, geographic region, target audience, business unit, campaign objective. Add these only if you report on them.

Work Backwards From the Report You Wish You Had

“I wish I could see which campaigns drive pipeline by product and region.” That sentence is where a taxonomy starts. The wish names your dimensions: channel, source, product, region. The dimensions decide which UTM parameters have to carry them: source, medium, campaign, plus extra fields where needed. The parameters need naming rules to keep their values clean.

Wish, dimensions, parameters, rules.

Work backwards from the report, never forwards from the parameters. Go the other way, starting from a list of parameter values, and you get the spreadsheet of approved UTMs nobody follows, because no value on it traces back to a report anyone asked for.

Open your analytics right now. Two questions.

What do I already see? Channels, sources, campaign names: whatever’s in your reports today. Be honest about how much of it you actually use.

What do I wish I could see but can’t? This is the list that matters. It usually starts like this:

Traffic says “facebook.” Organic or paid? No idea. And those are two different budgets defending themselves in two different meetings.
There’s a campaign name. Which product was it selling? The name won’t tell you.
Awareness campaigns and conversion campaigns, graded on the same curve, because nothing records what each one was for.
When did that campaign even run? Go find the spreadsheet. It’s an ex-colleague’s spreadsheet.

Every line on that list is a dimension you’re not capturing. That’s the whole diagnosis: your reports aren’t bad. They’re blind, one dimension at a time.

Follow Along: Stride Footwear

Throughout the guide we follow Stride Footwear, a fictional mid-size DTC shoe brand, as their tracking system takes shape.

Stride sells running and lifestyle shoes online and through retail partners. Their 4-person marketing team runs campaigns across email, paid social (Facebook and Instagram), Google Ads, organic social, and in-store QR codes. Their VP of Marketing, Dana, has a problem: the CEO keeps asking “Which channels actually drive shoe sales?” and Dana can’t answer confidently.

She opens GA4 and sees: facebook, Facebook, FB, fb as four separate sources. Campaign names like Spring Shoes, spring-shoes-2025, SS25-FB, and shoes_spring. An email source that’s sometimes newsletter, sometimes mailchimp, sometimes Email. Over 30% of traffic is “(direct) / (none).”

Dana’s Q2 board meeting is eight weeks away. She needs clean data by then. Her first step: figure out what she needs to see in that board report. She writes down four questions:

Which channel drives the most revenue: paid social, email, or paid search?

Which campaign converts best: the spring sale, the new product launch, or the evergreen brand campaign?

Are the in-store QR codes on retail displays worth the printing cost?

Is the weekly newsletter driving sales, or just traffic that doesn’t convert?

Four questions. Four dimensions: channel (medium), source, campaign, and online-vs-offline. That’s her starting point.

What Good Looks Like

Here’s the difference governance makes: the same report, with and without the rules.

TRY IT · Governance on / off

Same channel, same month, same spend. The only difference is whether people typed values by hand.

Session source	Sessions	Leads
Facebook	1,204	31
facebook	862	24
FB	391	9
fb	214	6
Face Book	12	0

One channel, five rows. Which number goes in the board deck?

Don’t Over-Engineer

Once you start thinking about what you could track, the list grows fast.

Be suspicious of that list.

Every dimension you add creates overhead: someone has to populate it correctly every time, maintain the allowed values, and make sense of it in reports.

Start with the dimensions that would genuinely change how you allocate budget or evaluate performance. For most organizations, that’s a short list:

Channel: the broad category (email, social, paid search, display, etc.)
Source: the specific platform or origin (facebook, google, newsletter)
Campaign: the specific initiative or promotion

Beyond those three, add only what your business specifically needs. If you sell multiple product lines, add product. If you operate across regions, add region. Don’t add dimensions on speculation. A dimension earns its place with a clear reporting use case, or it stays out.

Don’t Skip Goal and Date

Goal (Campaign Objective)

Most teams track where traffic comes from but not why the campaign exists. Add a goal dimension (awareness, lead generation, sign-up, sale) and you fix two problems at once.

First, everyone knows which metric to judge the campaign on. Lead generation gets measured on form fills. Sales gets measured on revenue. Without an explicit goal, the analyst has to guess what “success” meant for each campaign.

Second, in larger teams, the person building the UTMs is often not the person reading the report. A goal dimension is a message from one to the other: this campaign exists to do this, so measure it against that.

Date

Add a date dimension and you get analysis that’s otherwise surprisingly difficult. Does a campaign keep pulling traffic after the first week, or does it die quietly? That’s lasting power, and now you can measure it. Year-over-year comparison works directly, no spreadsheet archaeology required. And if you ever build automation around your tracking data, dates are essential.

On format: use YYYY_MM_DD in UTM values (based on ISO 8601, but with underscores so the date stays as one segment in structured naming). Any year-first, zero-padded format sorts correctly in spreadsheets, databases, and analytics tools, because alphabetical order and chronological order agree. Month-first and day-first formats fall apart the moment two years mix. YYYY_MM or YYYY_QN (e.g., 2025_Q2) is sometimes enough. The year always comes first.

You Can Always Add More Later

Tracking strategy is not a one-time decision. Start simple, prove the value, and layer in complexity as you earn it. (When the taxonomy needs to grow, Chapter 7 shows how to version it without breaking historical data.) Don’t build for hypothetical futures on day one.

Notice that nothing in this chapter mentioned UTM parameters. That’s intentional.

The dimensions you just defined (channel, source, campaign, product, region, objective) are your taxonomy. UTMs are how you deliver that taxonomy to your analytics platform, but they’re one carrier among several. The same dimensions govern your ad platform naming (Chapter 7b), your CRM fields (Chapter 10), and your warehouse schema (the For Data Teams playbook).

Get the dimensions right first. The parameters that carry them are a downstream decision.

If you remember one thing: Work backwards from the report, never forwards from the parameters. Channel, source, and campaign cover most teams. Add a fourth dimension only when a real report needs it.

Action Item: Open your analytics platform and look at your current campaign reports. Write down three things you wish you could see but can’t. Those gaps are the dimensions your taxonomy needs to capture.