Engagement metrics that actually mean something (and the ones that fool you)

Most engagement metrics are vanity dressed as insight. Time on page, scroll depth, clicks, sessions — they fill dashboards and feel meaningful, but a lot of them measure activity, not interest. Learning to tell the difference between a signal that predicts a sale and one that just looks busy is one of the highest-leverage skills in marketing measurement. Let's separate the two.

What behavioral data actually is

Behavioral data is the record of what people do, as opposed to who they are or what they say. Not their age or their survey answers — their actions. Which pages they viewed, how long they stayed, what they clicked, how far they scrolled, what they added to cart and abandoned. It's the most honest data you have, because it's what people did rather than what they claimed they'd do.

But "most honest" doesn't mean "self-explaining." A behavior is just an action until you interpret it, and the interpretation is where most measurement goes wrong. The same action can mean opposite things depending on context, which is why raw engagement numbers so often mislead.

The metrics that fool you: time on page and scroll depth

Take time on page. It sounds like interest — longer must mean more engaged, right? Not necessarily. Someone who spends four minutes on your page might be fascinated. Or they might be confused, hunting for a price you buried, or they left the tab open and walked away. High time on page can signal deep interest or deep frustration, and the number alone can't tell you which.

Scroll depth has the same problem. Scrolling all the way down can mean someone read every word — or that they were desperately searching for something they couldn't find and never did. A high number looks like a win and might be a symptom. These metrics aren't useless, but taken alone, out of context, they measure motion, not meaning.

Engagement signals worth trusting

So which signals actually matter? The ones tied to intent — actions that genuinely predict someone moving toward a purchase, rather than just moving. Adding to cart. Returning to the same product multiple times. Viewing pricing. Coming back within a few days. These are behaviors that, in context, correlate with buying, not just browsing.

The key phrase is "in context." A single add-to-cart from a first-time visitor means something different from the third visit to a pricing page by someone who's opened four of your emails. The signal isn't the action in isolation — it's the action as part of a pattern, from a person you can actually recognize across visits. Which is, again, why this comes back to identity: an engagement signal is only readable if you can connect it to the same person's other behavior over time.

Intent vs. behavior: the distinction that matters most

Here's the distinction that separates useful measurement from busy dashboards. Behavior is what someone did. Intent is what it means about where they're heading. Every behavior is a clue to intent, but no single behavior is intent itself — and confusing the two is how businesses misread their customers.

A pile of engagement — lots of clicks, long sessions, deep scrolls — can look like strong intent and actually be someone who can't find what they need. Meanwhile, genuine high intent might look quiet: a quick, decisive visit straight to the pricing page and out. Volume of activity is not the same as strength of intent. Reading intent correctly means looking at the pattern of behavior over time, from a recognized person, not tallying up raw engagement and hoping it points the right way.

How to actually use engagement data

• Distrust any single metric as a measure of interest. Time on page and scroll depth measure activity, and activity is ambiguous.
• Prioritize intent-linked signals. Cart adds, repeat product views, pricing visits, quick returns — these predict buying better than raw engagement.
• Always read behavior in context. The same action means different things from different people at different stages.
• Look at patterns, not moments. Intent shows up in a sequence of behavior over time, not in one impressive number.

Every one of those principles depends on a single prerequisite: being able to tie behaviors to the same person across visits and devices. A scroll or a cart-add is just noise until it's part of one person's story. That's the foundation Chapter is built on — it captures behavioral signals and ties them to a resolved identity, so engagement becomes a readable pattern of intent instead of a pile of ambiguous activity. For why recognizing the person comes first, see our guide to identity resolution.