Sally Gibson runs Dawleys, a contact center and fulfillment business in Ross-on-Wye, Herefordshire. When she joined Debra and me on Corey-osity Unleashed, she dropped a line about their company intranet I keep coming back to.
Engagement on it sits at 100%. Everyone reads it. Everyone joins in. And Sally reads the number as a culture check... one faster and more honest than any formal survey.
I spent decades building and running software. When I heard her say it, my first thought was "she's describing a production metric." My second thought was "and most leaders will wreck it within a month."
Here's why both thoughts hold.
Surveys Record What People Say. Usage Records What They Do.
Economists split preferences into two kinds. Stated preferences are what people tell you. Revealed preferences are what they do when nobody asks.
Your annual engagement survey collects stated preferences. Once a year, with your logo on top, from people who know their manager will see the results. I wrote about how badly the annual model performs in Your Annual Survey Has Been Lying to You. People answer for the audience.
Intranet usage collects revealed preferences. Nobody fills in a form to read the post about a colleague's new baby. Nobody gets a reminder email nagging them to congratulate the warehouse team on a record week. They do it, or they don't.
Sally's framing went further than reads. A busy intranet tells you people care about the company news. It also tells you people care about each other. A quiet one tells you the opposite, and it tells you sooner than a survey scheduled for next March.
Treat It Like a Production System
Every engineering team I've led had dashboards. The good ones followed a principle from Google's Site Reliability Engineering book. The chapter on monitoring distributed systems says your monitoring should answer two questions: what's broken, and why?
The "what's broken" part is the symptom. The "why" part is the cause.
Intranet engagement is a symptom metric. A falling read rate tells you something changed. It does not tell you what. A team lost a well-liked manager. A reorg left people unsure who to talk to. The posts got dull. Any of those fits the same falling line. The dashboard won't say. You have to go and find out.
Engineers learn a second lesson early. You watch the fleet, not one request. One slow call means nothing. A slow trend across every server means plenty. The same rule applies here. Team-level trends tell you about culture. One person's click history tells you about one person's Tuesday.
This is where most leaders go wrong.
The Moment You Zoom In, the Signal Dies
In 2022, Microsoft surveyed 20,006 workers across 11 countries for its Work Trend Index. 87% of employees said they were productive at work. 85% of leaders said hybrid work made it hard to have confidence their people were productive.
Microsoft named the gap "productivity paranoia." The report also warned about organizations tracking activity instead of impact. Employees often had no idea how or why their employer tracked them, which eroded trust and produced what Microsoft called "productivity theater."
Theater is the right word. Once people learn you count their clicks, they click. They open the post, close it, and move on. Your 100% stays at 100%, and it stops meaning anything. I covered this trap in Stop Chasing the Award... the moment a measure becomes a target, people serve the target.
The research on watching individuals is blunt.
A 2022 meta-analysis by Ravid, White, Tomczak and Miles pooled 94 independent samples covering 23,461 workers. They found no evidence electronic performance monitoring improves performance. They did find monitoring linked to higher worker stress, regardless of how the monitoring worked. The one bright spot: organizations which monitored more transparently and less invasively saw more positive attitudes from their people.
A second 2022 meta-analysis by Siegel, König and Lazar looked at 70 samples. Monitoring slightly lowered job satisfaction and slightly raised stress. Its relationship with performance came out at r = -0.01, which is to say none. And it showed a small positive link with counterproductive work behavior.
Read those two together. Watching individuals doesn't make them work harder. It makes them more stressed, a bit less happy, and a bit more likely to work against you.
So the same data gives you two different tools. Aggregated and shared, it's a culture thermometer. Broken out by name, it's surveillance with a friendlier logo.
How I'd Set It Up
If you run an intranet, a Slack workspace, a Teams tenant, or any internal platform with analytics, here's the version I'd stand behind.
Show everyone the same dashboard
Whatever leadership sees, the whole company sees. No private drill-down view for managers. The Ravid findings point straight at this: transparency changed how people felt about measurement. Hiding the numbers tells your team you don't trust them with the truth.
Aggregate to teams, never to names
Report by team or site, with a minimum group size so nobody gets singled out. Trailing four weeks beats daily snapshots. Daily numbers swing with holidays and deadlines and make you react to noise.
Count conversation, not consumption
Reads tell you people opened something. Replies, comments, and peer shout-outs tell you people care about each other, which was Sally's real point. Weight two-way traffic higher than one-way reads.
Watch the slope, not the score
A team at 60% and climbing worries me less than a team at 90% and sliding. The direction tells you whether something changed. The level mostly tells you about the team's habits.
Treat a drop as an invitation
When a line falls, book a coffee, not a review. Ask what's changed for them lately. You're investigating a symptom, so go find the cause with the people who know it.
Keep it out of performance reviews
Forever. The moment intranet activity shows up in anyone's appraisal, you've turned a thermometer into a target, and the theater starts.
A Dashboard Won't Fix a Bad Boss
My research for Step It Up HR found 99.5% of survey respondents said they'd had one or more types of bad boss. The Micromanager sits high on the list. I've written before about two separate surveys landing on the same villain.
Give a micromanager a per-person activity dashboard and you haven't given them insight. You've given them a better magnifying glass.
Give a decent manager a team-level trend line and a habit of asking questions, and you've given them early warning. Same data. Different leader.
A Smoke Alarm, Not a Fire Brigade
Here's how I think about Sally's 100%.
A smoke alarm doesn't put out fires. It doesn't tell you which room is burning or why. It tells you to get up and go look. Nobody sane wires a smoke alarm to report which family member burned the toast.
Your intranet numbers work the same way. They tell you where to look and when. Then you go and talk to people, because the "why" lives in conversations, not in logs.
Sally doesn't rely on the dashboard instead of talking to her team. She relies on it to know when to talk to them sooner.
So here's my challenge for this week. Open whatever analytics your internal platform offers. Look at the team-level trend for the last three months. Then ask yourself one question:
If this line dropped tomorrow, whose coffee would I book first... and why haven't I booked it already?