Atiba de Souza.Start a conversation
WritingMarketing

Nobody's Signing Up. Why Are You Tracking Clicks?

By Atiba de Souza

An editorial cover image for a piece on Nobody's Signing Up. Why Are You Tracking Clicks?. The cover reads: MARKETING Nobody's Signing Up. Why Are You Tracking Clicks?

Two smart people, one wrong question

Two members of my marketing team came to me a few months back with a request. They wanted tracking codes added to the webclass invitations. Each of them would get their own code, so we could see which of them was driving the most clicks.

Good idea. Clean execution. Real ownership. Everything you say you want from a marketing team.

I turned it down.

Not because tracking is wrong. Because nobody was signing up. My words to them were not gentle: "No one's signing up. So why are we worried about tracking where people are coming from if we don't have anyone signing up? What is it gonna prove?" Then the question that actually mattered: "Where's the actual constraint?"

They had brought me a fix for a stage that was not broken. If I had said yes, we would have built a beautiful measurement system on top of a page nobody converted on, and we would have learned exactly one thing: which of two talented people was better at generating traffic to a dead end.

Sign-ups stayed at zero. That is the number that decides what to work on, and it was not on the dashboard.

That is not marketing. That is scorekeeping at the wrong end of the field.

The question every channel decision skips

The question is not "which channel works?"

It is: what changes in the system if this works?

Run the tracking request through it. Suppose it works perfectly. Suppose we know, to the click, which person drove more traffic. What changes? Sign-ups stay at zero. Nothing in the system moves. The idea was good, the timing was wrong, and the timing is the whole thing.

A good idea aimed at a non-constraint is still waste. Test every idea by asking what changes in the system if it works.

That is the test. Not "is this a good idea?" It probably is. Ideas that survive a team meeting are usually good ideas. The question is whether it touches the thing that is actually broken.

Constraint: the one stage of the customer journey that is actually losing people right now. Not the stage that is easiest to see, not the stage your team is excited about, and not the stage your tools already happen to measure.

One sign-up is one data point

Here is the analogy I keep coming back to. One success is one data point, and one data point says nothing about the channel that produced it.

If one person signs up, you can learn their name. You cannot learn which channel works. Attribution needs volume. You cannot divide up an audience you never had. Working out where the leads came from, before there are leads, is not analysis. It is a very organized way of avoiding the actual problem.

This is where channel arguments go sideways. Somebody brings a fix that is real and useful and aimed at a stage that is not losing anyone, and the room agrees because the fix is smart. Smart is not the bar. Relevant is the bar.

Where this lives in the 7 Practices of Effective Marketing

The framework I teach is the 7 Practices of Effective Marketing. This decision sits inside three of them, and the order is the point. Run them in sequence on the webclass request and the answer takes care of itself.

Practice 1, The Win. Every game has a way to score, and everyone playing knows what scoring looks like. Marketing is no different. Before any work begins, the win has to be defined in terms that can be measured, because without a defined win there is no way to tell whether a project succeeded, no way to course-correct, and no way for anyone, human or AI, to evaluate the work being produced. For the webclass, the Win was sign-ups. Not clicks. Not engagement. Sign-ups. The moment the Win is sign-ups, the tracking code has to answer to sign-ups, and it cannot.

Practice 2, Thinking Steps Not Tactics. The instinct in marketing is to jump straight to tactics: funnels, ads, launches, campaigns, tracking codes. That is backwards. Before any tactic is considered, the logical steps required to reach the Win have to be mapped out in order. For the webclass that means the invitation, the landing page, the sign-up form, the confirmation, the attendance, each one dependent on the one before it. Tactics are how you execute a step. They are not a substitute for the step. The tracking code is a tactic, and until the step it serves has a Win attached, it serves nothing.

Practice 3, Narrow the Focus. At any moment there is usually more than one important thing to do. Narrowing means choosing the one that matters now and letting the others wait, out loud, on purpose. On the webclass, the constraint was downstream of the click, at the sign-up itself. So the tracking code got parked. Not killed. Parked, because it answers a real question that matters once the constraint moves. It just was not the question in front of us.

Run that sequence and the answer is never in dispute. Skip the sequence and the tracking code looks like progress.

No

Yes

Practice 1: Define the Win. For the webclass, sign-ups.

Practice 2: Map the ordered steps from here to the Win

Practice 3: Name the constraint, the step actually losing people

Does the idea touch the constraint?

Park it. Good idea, wrong stage, still waste.

Now choose the tactic or channel

Measure against the Win, not what is easy to count

The trap I built myself

Now the part where I got it wrong.

I own marketing for TrueMD, a three-founder venture. In July 2026 my team was building its LinkedIn presence and assets, and work started getting dropped. Then assets started appearing that made no sense to me. Off-target, off-message, work that answered a question nobody had asked.

I traced it back, and here is the admission: I made a major blunder. Most of my team had never been told what TrueMD actually is. Only a certain subset knew. I had assumed everyone did.

Sit with that for a second. I teach the 7 Practices. I ask other people where the actual constraint is. And the constraint in my own lane was that the people doing the marketing could not describe what we were marketing. No channel decision can be made by a team that does not share a definition. All marketing is communication, and I had not communicated.

The fix is not a better dashboard or a sharper channel strategy. Write down what TrueMD is. Then confirm every person working on it has read it before work starts. Not a subset. Every person.

That is the turn. The sophisticated answer was writing the definition down and making sure the whole team read it. I would rather hand you that lesson than let you buy it the way I did.

Inconsistent vs consistent

InconsistentConsistent
Pick the channel, then find a way to justify itName the Win, then find the stage that is losing people
Track clicks because clicks are availableTrack the stage that decides the Win
Split credit between teammatesSplit nothing until there is volume to split
Ask "is this a good idea?"Ask "what changes in the system if this works?"
Assign work before the team shares a definitionWrite the definition, confirm everyone read it, then assign
Measure what the tools measureMeasure what the Win needs

The reason we skip it

Why does a smart team skip this question? Because asking it costs something, and the cost is emotional, not analytical.

It costs you the feeling of progress. A tracking code is a deliverable. "Where is the actual constraint?" is a question, and questions do not show up on a status report.

It costs you the comfort of the dashboard. Numbers feel like control. Saying "we do not have enough volume for these numbers to mean anything" feels like handing over your instruments in fog.

It costs you a yes. I had to tell two people who did good work that their good work was aimed at the wrong stage. That is not fun, and if you are the kind of leader who says yes to keep the room happy, this practice will sting.

And it costs you the story you tell about yourself. If the constraint is further along the journey, then the work you have been pouring yourself into is not the work that matters. That is a hard sentence to say about your own last quarter.

Name the fear precisely, because it is not the fear of bad tracking. The fear is that if you stop counting clicks, someone above you will ask which channel works and you will have nothing to show. So you measure the thing you can measure instead of the thing that decides.

Here is the rewiring. Measurement is not a comfort object. It is a diagnostic. A diagnostic that cannot change a decision is decoration.

When not to run this

Do not turn this into a rule you hide behind.

If you already have volume and a funnel you understand, attribution is a real question with a real answer, and tracking is exactly the right tool. The failure in my story was not tracking. It was tracking before there was anything to track.

And do not use "where is the constraint?" as permission to never ship. At some point you have to test. Narrowing focus is not paralysis. Pick the constraint, run the test, read it against the Win, move.

What to do Monday

Monday morning, before your next channel conversation, write two things down: the Win in one measurable sentence, and the ordered steps from here to that Win with the constraint named. That is the prep. It takes an hour. It is not the hard part.

The hard part is the next tracking request, because there will be one, and it will come from someone good.

When it lands, do not answer the request. Ask the question: what changes in the system if this works? Say it plainly and wait. If the honest answer is "we would know more," say that, and then do the thing most leaders skip. Thank them for the instinct, and separate the instinct from the aim. "You are right to want to know what works. This will not tell us, because not enough people are getting far enough for the numbers to mean anything." You are not turning down their idea. You are turning down the target. Say the difference out loud, because the gap between "no" and "not there" is the whole relationship.

Then handle the cost you just created. Those two people on my team did real work. They designed the codes, planned the implementation, and brought it to me with ownership. When I said no, that effort did not vanish. It went somewhere, and where it goes is your job. Point it at the constraint. Give them the stage that is actually losing people and let them own that instead. The skill that built the tracking code is the skill that fixes the real funnel. Do not waste it, and do not let them believe you wasted it.

Then hold the no when the dashboard is demanded. Someone above you will ask which channel works, and your answer cannot be a number you do not have. It has to be a sentence about the constraint, the Win, and what you are doing about it, delivered with enough steadiness that it does not sound like an excuse. That is the rewiring in practice. You stop treating measurement as a comfort object and start treating it as a diagnostic. A diagnostic that cannot change a decision is decoration. Say that to yourself before you say it in the room.

And before any of it, write down what your venture is, and confirm every person working on it has read it before work starts. That one is mine to own. Do not make it yours the way I made it mine.

The next time someone brings you a fix, do not ask whether it is a good idea. Ask where the actual constraint is. The answer tells you whether you have work in front of you or a hobby in front of you.

Questions

What people ask next

Okay, but how do I actually tell which stage is the real constraint instead of just picking the one that's easiest to see or the one my team is already excited about?
Map the ordered steps from wherever your prospect enters to the Win, then look for the step where people are actually dropping out right now, not the step your tools happen to track or the one your team is pumped to fix. In the webclass case that meant ignoring click data entirely and looking straight at sign-ups, because that was the number stuck at zero. The test is simple: run the idea through 'what changes in the system if this works,' and if the answer is nothing, you're looking at the wrong stage.
If I keep sending good ideas back because they're aimed at the wrong stage, won't my team eventually stop bringing me ideas at all?
That risk is real if you reject the idea and the person in the same breath, but the article's move is to park, not kill. Tell the team explicitly that the idea is good and will matter once the constraint moves, and that the test being applied is the same one every idea has to pass, not a personal judgment on their work. The bar is relevant, not smart, and once people understand that's the actual standard, being told 'wrong stage, not now' stops feeling like rejection.
How many sign-ups or conversions do I actually need before it's even worth looking at which channel or person is driving them?
The article doesn't give a magic number, and the honest answer is it depends on having enough volume to meaningfully divide an audience. What it does say clearly is that one success is one data point: if one person signs up, you learn their name, not which channel works. The practical rule is to wait until the constraint itself (sign-ups, in that example) is no longer stuck at zero before spending effort attributing where those sign-ups came from.
Once the constraint moves downstream, what's the signal that tells me it's time to go back and unpark the ideas I shelved?
The signal is that the stage the parked idea was built for is no longer the one losing people, and a new stage downstream has become the bottleneck instead. Re-run the same test on the parked idea: ask what changes in the system if it works now. If the answer has flipped from 'nothing' to 'something moves,' that's your cue to unpark it.
What stops 'write down the definition and make sure everyone reads it' from turning into just another document nobody actually opens?
The fix described isn't just writing the document and distributing it, it's confirming every person working on it has actually read it before any work starts, not a subset, everyone. That confirmation step is what turns it from a passive file into a gate: no one gets assigned work until they've demonstrated they understand the shared definition. Skip that gate and you're back to the exact blunder the article admits to making.