Which Events Deserve to Be Conversions (And What the Rest Cost You)

A.It makes the number uncountable

Conversions are summed. If you have marked signup, demo request, pricing page view and newsletter subscribe as conversions, a single motivated visitor who does all four contributes four conversions, and a visitor who signs up and nothing else contributes one.

The conversion rate that comes out of this is not wrong so much as meaningless. It is answering “how many labelled things happened per session”, which is not a question anybody has. Two channels can be compared on it only if they produce the same mix of labelled things, and the entire point of comparing channels is that they do not.

B.It teaches your ad platforms the wrong lesson

This is the part that costs money rather than clarity. Modern ad bidding is a machine that finds more people like the ones who converted. It is very good at this, which is precisely the problem when the definition is loose.

Import a newsletter subscription as a conversion and the bidder will faithfully go and find people who subscribe to newsletters. Those people exist in enormous numbers, they are cheap to reach, and a meaningful share of them will never buy anything. Your cost per conversion will fall, your reporting will improve, and your revenue will not move. It is one of the few failure modes in marketing that looks like success the entire time it is happening.

The bidder does exactly what you told it. The instruction was wrong.

C.It hides the event that mattered

The third cost is attention. A report listing eleven conversions is a report nobody reads past the first two rows. The purchase event, the one that pays for the company, sits alphabetically between pricing_page_view and scroll_90, and gets the same visual weight as both.

A.Would you move budget because it moved

The first condition does most of the work. If this number went up 30% next month, would anyone reallocate spend? If the honest answer is that you would want to know why but you would not act, it is a diagnostic, and diagnostics belong in funnels.

Apply it to the usual suspects. A purchase: yes, obviously. A trial signup for a product with a known trial-to-paid rate: yes, because it converts to expected revenue. A pricing page view: no. It is a strong intent signal and you would never move budget on it alone, because you can double pricing page views with a nav change.

B.Is it hard enough to be honest

The second condition is whether the event is cheap to trigger accidentally. Anything based on scroll depth, time on page, or a pageview fails here. These fire for bots, for mis-clicks, for people who left a tab open, and they can be inflated by a design change that has nothing to do with intent.

A conversion should require the person to have decided something. Form submitted, payment completed, account created, meeting booked. The threshold is deliberate action, not presence.

C.Does it have a known value

The third condition is whether you can attach a number to it, even a rough one. A purchase has an order value. A trial signup has an expected value if you know your trial-to-paid rate and your average contract value. A demo request has one if you know your close rate.

An event with no value cannot be traded off against another event, which means it cannot support the budget decision the label exists to enable. Our guide to calculating LTV covers where those numbers come from.

A.The primary, and only one

Pick the single event that most directly represents money changing hands or committing to change hands. For e-commerce it is the purchase. For self-serve SaaS it is the account created, or the first payment if your trial-to-paid rate is unstable. For sales-led B2B it is the qualified meeting booked.

That is your conversion. If a second one is genuinely necessary, because you run two distinct motions in one account, keep both and stop there.

B.Everything else becomes a funnel

The objection to demoting events is always the same: we still need to see them. Of course you do. That is what a funnel report is, and it reads better as a funnel than as eight competing conversion lines, because a funnel preserves the order and the drop-off between steps while a list of conversions destroys both.

Newsletter subscribes, pricing views, add-to-carts and video plays are all excellent diagnostic steps. They tell you where attention is lost. They just should not be competing with revenue for the same label. Our guide to micro-funnels covers building those, and why GA4 funnels and revenue reports disagree covers the trap underneath them.

C.What a self-configuring schema does with this

The reason this decision gets made badly is that it gets made early, by whoever is implementing tracking, before anybody knows which events will matter. It is a judgement call presented as a setup step, and the cost of getting it wrong surfaces two quarters later in a bidding algorithm.

Kissmetrics does it in two passes. Point it at your site and it works out what kind of business you run and what is worth capturing for one. Then, once events are actually arriving, it gives each event a role from a fixed vocabulary, conversion among them, and nominates the one that looks like your primary conversion. The bet is not that the inference is always right. It is that reviewing a draft where the roles are already assigned is a very different task from being handed an empty conversions list on day one and asked to guess.

The correction matters as much as the draft, which is why the nomination is labelled as a suggestion and has a picker beside it. If it ever decides your newsletter is the primary conversion because it looks like one from the outside, you want to see that in an afternoon rather than in next quarter's cost per acquisition.