UTMs, Promo Codes and Vanity URLs: Instrumenting a Creator Campaign

Most creator campaigns are instrumented the evening before launch, by whoever is free, and the consequences show up a quarter later when nobody can answer which creator produced the customers.

The measurement framework — what to count and against what — is a separate question. This is the plumbing: the naming, the mechanics, and the specific places attribution leaks in creator campaigns that do not leak elsewhere.

Decide the convention first, and freeze it

UTM parameters are simple. Keeping them consistent across twenty creators, three platforms and four quarters is not, and inconsistency is what makes a campaign unanalyzable.

A scheme that holds up:

Parameter	Value	Why
utm_source	Platform — youtube, x, linkedin, newsletter	Lets you compare platforms
utm_medium	Always creator	One medium rolls every placement into one bucket
utm_campaign	Campaign name — launch-q4-2026	Groups a wave
utm_content	The creator — janedoe	Per-creator reporting, one dimension
utm_term	Optional — format, e.g. dedicated-video	Compare formats without a second campaign

So a placement looks like:

https://www.example.com/
  ?utm_source=youtube
  &utm_medium=creator
  &utm_campaign=launch-q4-2026
  &utm_content=janedoe
  &utm_term=dedicated-video

Four rules that matter more than the scheme itself:

• Lowercase, hyphens, no spaces. Analytics tools treat YouTube and youtube as different sources, and you will find out in the report.
• utm_medium is fixed forever. The moment one person writes influencer and another writes creator, your channel report splits in two and stays split.
• One creator, one utm_content value, for life. Reuse the same handle across campaigns and you get longitudinal data on that creator for free.
• Generate them from a sheet, not by hand. A single column of formulas, one row per creator, copied into the brief. Hand-built URLs produce typos that silently route conversions to nowhere.

Write the convention down in the campaign doc. The person instrumenting the next campaign will not be you.

Links, codes and spoken URLs are three different instruments

The most common instrumentation failure in creator work is using a clickable-link mechanism in a format where nobody clicks.

UTM’d links work where the link is clickable and adjacent to the content: video descriptions, newsletter placements, X posts, blog embeds. They are precise and they are also fragile — a viewer who watches on a TV, or on mobile without opening the description, is invisible to them.

Vanity URLs are for anything spoken. example.com/jane is sayable, memorable, typeable from a podcast, and it can redirect to the full UTM’d URL server-side so the analytics stay clean. If a creator will say the URL out loud — and for video and audio they should — this is the only mechanism that captures it. Keep them short, all lowercase, and avoid characters that are ambiguous when spoken.

Promo codes survive everything a link does not. A code gets screenshotted, pasted into a group chat, remembered for a week, and typed at checkout by someone who never clicked anything. That resilience is exactly the dark social path that link tracking loses.

Their limitation is structural: a code only fires at a checkout. If your funnel is a free signup with payment three weeks later, the code records a fraction of the people the placement actually sent, and it records them late.

The practical answer for most self-serve products is all three: a vanity URL that redirects with UTMs attached, spoken aloud and in the description, plus a code for the paid step. They measure different populations and the overlap is informative rather than double-counted, as long as you decide in advance which one is the headline number.

One more: a dedicated landing page per major creator. Worth it for your largest few placements. It lets you mirror the creator’s framing, and page views give you an unambiguous top-of-funnel count that no attribution setting can argue with. Not worth the maintenance for a long tail of small placements.

Capture attribution at signup, not in the session

This is the failure that quietly invalidates the most reporting, and it is a product decision rather than a marketing one.

A B2B buyer watches a video on Tuesday, thinks about it, searches your name the following Monday, and signs up. If your attribution lives in a session or a short-lived cookie, that customer is recorded as organic search. The creator gets nothing. Every creator gets nothing, systematically, because considered purchases always involve a gap.

What to do instead:

1. Persist the first-touch parameters in durable storage on landing, and keep last-touch separately. First touch tells you who created the demand; last touch tells you who closed it. Both are true and they answer different questions.
2. Write them onto the account record at signup, server-side. Not into an analytics tool only — into your own database, on the user or organization row. Analytics tools change, retention windows expire, and cross-domain tracking breaks; a column on your users table does not.
3. Carry it through to revenue. The number that matters is cost per customer, which means the attribution has to survive from the click to the invoice. If it stops at the signup event, you will be comparing creators on signups forever.

If you take one thing from this article, take this section. The UTM convention is hygiene; storing attribution on the account is what makes the whole channel measurable.

Designing the “how did you hear about us” field

Crude, unscientific, and consistently the most accurate instrument you have for creator attribution.

Design notes that materially change its value:

• Free text, not a dropdown. A dropdown gives you the categories you already guessed. Free text gives you creator names, and creator names are the entire point.
• Ask at signup, while the memory is fresh, not in a later survey.
• Optional, and placed after the account is created so it does not depress conversion.
• Read it monthly and normalize into a rough tally. Someone typing a creator’s handle, a channel name or “that video about the queue migration” is telling you something your dashboard cannot.

Expect a 20–40% completion rate, and treat it as directional. The signal you are looking for is relative: which names appear, and how the volume moves when a campaign runs.

Measuring what tracking structurally cannot see

Tracked links are a floor. Three methods find the rest, in ascending order of rigor.

Branded search and direct traffic lift. Establish a baseline for searches on your product name for four weeks before the campaign, then watch the change during and after. Creator content produces name searches almost mechanically — someone watches, does not click, and looks you up later. If branded search rises 30% while tracked links show 40 signups, you now know roughly how much of an underestimate the 40 is. This is free in Search Console and almost nobody looks at it.

Time-series against a known baseline. Total signups per day, with campaign dates marked. Blunt, and it catches effects that per-placement attribution cannot decompose.

A geographic holdout. Run creators in some regions and not in others, keeping every other channel constant, then compare total signups rather than attributed ones. This is the only clean incrementality read most companies can afford, and it survives every attribution argument because it does not depend on attribution at all. It requires the discipline to leave money on the table in the holdout region, which is why it is rarely done and worth doing once.

The instrumentation checklist

Before any placement goes live:

• The UTM convention is written down, and every creator’s link was generated from the same sheet.
• Every link has been clicked and verified — parameters intact after every redirect, and after the platform’s own link wrapper.
• A vanity URL exists for anything that will be spoken, and it redirects with the UTMs attached.
• A unique code exists per creator if there is a checkout, and it is live in billing before the content publishes.
• First-touch and last-touch attribution are persisted to the account record at signup.
• The “how did you hear about us” field is live and free-text.
• A four-week baseline for branded search and direct traffic has been recorded.
• The brief tells the creator exactly where the link and code must appear: spoken, in the description, and pinned in the first comment.

That last line is where more attribution is lost than in any technical detail. A perfectly instrumented link mentioned once at the bottom of a description is an untracked placement.

Common mistakes

• Changing the convention mid-campaign. Two mediums, two channel reports, no comparison.
• A code that goes live after the video does. Viewers try it, it fails, and the ones who care enough to try are the ones you lost.
• Only measuring last click. Search takes the credit for demand creators generated — the mechanics of that are in creator marketing versus paid search.
• Reporting in week one. Creator content compounds for months; an early number is a fraction of the eventual one and is how a working channel gets cut.
• Comparing a click-only creator number to a paid platform’s view-through number. Different definitions, and the comparison is meaningless without saying so.

Where ActiVibe fits

We instrument every placement before it goes live — link, vanity URL and code per creator, verified end to end — and report on cost per signup and per customer per creator rather than on reach. The unglamorous half of this channel is the half that decides whether anyone gets to run it again next quarter.

Model the expected numbers first with the ROI estimator, or get a free GTM strategy that includes the tracking plan alongside the roster.