User behavior before upgrade

Most self-serve accounts don't upgrade out of nowhere. Something changes in how they use the product first: a team invites more people, they hit a limit, they start using a feature they'd ignored for weeks. Spotting that shift before the upgrade happens, instead of after, is what turns a billing event into something you can actually act on.

Why upgrade timing is a behavior problem, not a billing problem

By the time an account upgrades in your billing system, the decision is already made. If you want to influence that decision, or simply understand your own funnel well enough to forecast it, you need to look earlier: at what the account was doing in the days or weeks before.

This is different from asking "is this account engaged" in general. A team can be engaged for months without being close to upgrading. What matters here is a change: usage crossing a threshold it hadn't crossed before, a new user type showing up, a feature getting adopted that wasn't part of the account's original habit.

What Funnelsight shows you

Funnelsight keeps product usage and CRM data in the same funnel view, across activation, retention, and expansion. That expansion axis is where pre-upgrade behavior lives: seat growth, feature adoption patterns, and usage frequency, tracked over time rather than as a single snapshot.

You define the thresholds that matter for your product (a seat count, a feature touch, a frequency floor), and Funnelsight applies them consistently across every account. If you're already using expansion revenue signal tracking to flag accounts likely to grow, this is the same underlying data, viewed from the angle of "what changed right before" rather than "what does this account look like now".

Data refresh is configurable, not continuous. For most teams watching pre-upgrade behavior, a daily or twice-daily refresh is enough to catch a pattern early without needing a live stream.

Start with one or two signals, not a full model

You don't need to map every possible pre-upgrade behavior on day one. Pick the one or two that are most obviously tied to upgrades in your product, seat count is a common starting point for team-based tools, and watch those first. A spreadsheet export can work for a first pass if you're not ready to connect a live source yet. Add more signals once the first ones prove useful, rather than trying to cover every scenario from the start.

A concrete example

Say your product has a free tier capped at three projects. An account that goes from one project to three within a week, and then invites a fourth teammate, is behaving differently from an account that's been sitting at one project for two months. Flagging that shift, rather than waiting for the account to hit the cap and get a billing prompt, gives your team (marketing or sales) a window to reach out with context, not a generic upgrade nudge.