Power User

For a long time the only way to describe the people using a product was an average. Twelve minutes a day, four documents per person. Those numbers were easy to compute and they described nobody. Half the accounts signed up, looked around once, and never came back. A few dozen lived in the product for hours every day. Average the two groups together and you get an imaginary person who matches neither.

Meetings kept stalling at the same spot because of it. Someone would ask whether anyone actually used a feature, the answer would come back as "three percent," and the conversation would end there. Nobody could say whether those three percent had clicked it once out of curiosity or were the teams paying every month. A bare percentage could not tell the room whether cutting the feature would cost nothing at all or cost them their best accounts.

Power user is the name for the layer at the deep end of that lump. The idea is to stop flattening people into one average and start splitting them by how much of the product they actually use.

A power user is defined by one threshold the product writes down for itself. The team decides in advance which behaviors, at what frequency and at what breadth, qualify an account for the group, and then re-draws the list every week from the event data.

Thresholds usually combine three things.

1. Frequency — how often someone shows up, written as a number of days, such as five or more days a week. Some teams count sessions instead. A session is one visit bundled from arrival to departure, and it's normally treated as over after 30 minutes without activity.
2. Depth — how much of the core action someone performs. At Pagedeck that means how many documents an account created and how many comments it left.
3. Breadth — how many distinct features someone touches. An account that only writes documents and one that writes, shares, comments, and builds templates sit in different layers even when they spend the same number of hours. Feature adoption rate measures the share of people who tried one feature; breadth flips that around and counts the features one person has tried.

Use only one of the three and the wrong group falls out. Frequency alone catches accounts that open the app daily, glance at notifications, and leave. Depth alone catches the account that spent a single day moving 30 old documents in and then went quiet.

Where the line goes is a local decision, not something to import. There are two common shapes. A percentile cut ranks every account by usage and takes the top five percent. An absolute cut names fixed numbers instead: Pagedeck counts an account as a power user when it shows up five or more days a week and has three or more documents it co-edits with someone else. The percentile version keeps the group the same size as the product grows. The absolute version keeps the bar still, which makes last month and this month directly comparable.

The group earns its keep in the meeting where a team decides what to build next. Four uses come up most often.

1. Feature calls — when deciding whether to cut or keep something, put the overall usage share next to the share inside this group. A feature used by three percent of everyone but 68 percent of power users isn't a feature nobody wants; it may be the reason the longest-staying accounts stay.
2. Testing early — open a new feature to this group before opening it to everyone. They have used the product long enough to notice what changed, and they can describe where it snags in words a team can act on. There's overlap with early adopters, the people who reach for new things first, but the two aren't the same. Early adopter describes a disposition; power user describes accumulated usage.
3. Reading other numbers — monthly active users (MAU) counts the people who used the product at least once in a month. If MAU climbs from 8,400 to 12,000 while the power user count holds flat, headcount grew and depth didn't. Stickiness, the share of a month's users who show up daily, gets traced back through this layer when it drops.
4. Keeping and growing accounts — when this group starts leaving, the signal arrives weeks before retention, the measure of how reliably past users come back, actually bends. Paid upgrades and plan increases tend to start here too, which is why sales asks for the list separately.

"Overall it's three percent, but 68 percent of power users touch it. Cut it and they go first."

Countering a proposal to kill a feature on its headline percentage with a different piece of evidence.

"Next week's beta goes to 200 power users only."

Show it first to people who will notice problems and explain them, fix what comes back, then widen it.

"Let's agree on our power user definition. Every team is counting it differently."

The threshold was never written down, so the same name now refers to several different groups.

At Pagedeck, Mia re-pulls this list every quarter. Filtering for accounts that show up five or more days a week with at least three co-edited documents leaves about 200 of them, and she lines up per-feature usage inside that group against the whole base. Features where the two numbers diverge sharply become candidates for the next quarter.

The most common mistake is reading the group as a specification: build what power users ask for and the product gets better. It sounds plausible because they're vocal and they know the product well enough to make concrete requests. But these are people who finished adapting to the product long ago, so their requests lean toward finer settings and more controls. Follow them literally and the screen gets busier for everyone arriving today. Six months later power user satisfaction is up while half of each week's new accounts leave before the second week. Treat what this group asks for as a signal about where the product binds, not as an instruction about what to build.

Reading heavy use as satisfaction is the other one. Usage and satisfaction overlap often enough that separating them feels unnecessary. Work tools break that pattern, though, because some accounts keep logging in only because nothing else is available. An account that appears every single day handing out the lowest possible score on net promoter score (NPS), which asks how likely someone is to recommend the product, is a real and ordinary outcome. Trust usage alone and this is the layer that leaves first the month a credible competitor ships.