Key Takeaways
The dashboard took a quarter to build. It had 24 tiles: health score gauges, NPS trend, logins, tickets, renewals, a pie chart of accounts by segment. It went up on a screen in the office and into the Monday meeting. By month two, the meeting skipped straight to the renewals list. By month four, nobody opened it between meetings. Carmen asked the team which tile had changed a decision that quarter. Nobody could name one. For usage signals worth tiling, see product adoption metrics that predict renewal.
That's common, and it isn't a design failure you can fix with better colours. Most customer success dashboard guides list metrics and templates. Few ask whether dashboards change what people do. The research on that question comes mostly from healthcare, where dashboards and performance feedback have been tested in controlled trials. This post covers what it says, and a five-question test you can run on every tile.
What a dashboard is for
The most cited definition is still the standard. A dashboard is “a visual display of the most important information needed to achieve one or more objectives; consolidated and arranged on a single screen so the information can be monitored at a glance.” Note the words “most important” and “single screen.” A 24-tile dashboard usually fails the first test before anyone reads it.
Design guidance is thinner than you'd expect. A 2022 review of 144 dashboards found “there is surprisingly little design guidance for dashboards” and called evaluation “a current gap in the literature.” A 2025 review of 89 dashboards found only “47% (n=42) of dashboards were still accessible or active.” Many dashboards are built, used for a while and then abandoned.
What the feedback research says
A dashboard is a form of feedback: it shows people how they're doing. Feedback has been studied for decades.
The biggest review covered 607 effect sizes. It found feedback “improved performance on average (d = .41) but that over 1/3 of the FIs decreased performance.” It also found why some backfired: “FI effectiveness decreases as attention moves up the hierarchy closer to the self and away from the task.” Leaderboards and rankings pull attention toward the person. Task-level signals keep it on the work.
Healthcare has tested this in controlled trials. A Cochrane review of 140 trials found that showing professionals their performance data led to “a 4.3% (interquartile range (IQR) 0.5% to 16%) absolute increase” in following recommended practice. That's a small effect on average, with a wide range.
The same review found what made the difference. Feedback “may be more effective when baseline performance is low, the source is a supervisor or colleague, it is provided more than once, it is delivered in both verbal and written formats, and when it includes both explicit targets and an action plan.” A follow-up analysis found the average effect “became stable in 2003”. New trials kept testing display without testing those design factors.
Studies of dashboards specifically point the same way. One review found that “in contexts where dashboards were easily accessible to clinicians (such as in the form of a screen saver) their use was associated with improved care processes.” A 2022 review of 11 trials found only “limited evidence” of improvement, and mainly where dashboards were part of a wider intervention.
This is clinical evidence, and we're applying it by analogy. There's no equivalent research on CS dashboards. But the pattern is consistent across settings: display alone does little. Display with a target, a plan, a named person and repetition does more.
The five-question tile test
Run every tile on your customer success dashboard through five questions. Each comes from the research above.
| Question | Why it matters | What to change |
|---|
| 1. Compared to what? | A number without a target or history doesn't prompt action | Add a target line or last period |
| 2. What do we do if it moves? | Feedback with an action plan works better | Write the action next to the tile |
| 3. Who says it out loud, and how often? | Feedback from a named person, repeated, works better | Name an owner and a cadence |
| 4. Is it read by position? | People misjudge angles and areas more than positions | Replace gauges and pies with bars or lines |
| 5. Is it the goal, or a stand-in? | Teams start managing the metric instead of the goal | Name the goal the tile serves |
Sources: Perceptual Edge, 2006; Cochrane review, 2012; Stanford perception study, 2010; HBR, 2019. The “what to change” column is our recommendation.
Question 1: Compared to what?
A widely read guide on dashboard design puts it directly: “Compared to what? Is this good or bad? … The right context for the key measures makes the difference between numbers that just sit there on the screen and those that enlighten and inspire action.” A health score of 72 means nothing on its own. A score of 72 against a target of 80 and last month's 78 is a prompt.
Question 2: What do we do if it moves?
This is the question most tiles fail. If a renewal-risk count rises by five, who calls which accounts, and by when? The Cochrane review found feedback works better when it includes “explicit targets and an action plan.” Write the action next to the tile. If nobody can write one, the tile is information, not a prompt.
Question 3: Who says it out loud, and how often?
The same review found feedback was more effective when the source was a supervisor or colleague, it was provided more than once, and it was delivered in both verbal and written formats. For a CS team, that means a named owner who reads the tile aloud in a regular meeting and says what they did about it. The feedback review's warning applies here too: keep the conversation on the task, not the person. A tile that ranks CSMs against each other pulls attention away from the accounts.
Question 4: Is it read by position?
Gauges and pies are popular on CS dashboards. They're also hard to read accurately. A Stanford replication of classic perception studies found “area to perform worse than angle, and both to be significantly worse than position.” A bar chart or a line with a target is easier to read correctly.
Question 5: Is it the goal, or a stand-in?
Harvard Business Review calls the risk surrogation: “employees often mentally replace it with the hard metrics meant to assess whether the organization is succeeding at it.” Health scores are the classic example. Gainsight's 2023 index found “Customer Health Scores were the number one non-revenue measure of CS organizations in North America, with 60%.” TSIA warns that “a customer may appear ‘green’ or ‘healthy’ on your dashboard, while behind the scenes, churn risk is already rising.” A green tile isn't a renewed customer.
Why most teams can't build the single view
Even a well-designed dashboard needs data. Most CS teams pull it from many systems. MuleSoft's 2025 benchmark of 1,050 IT leaders found organizations use “897 applications, on average” and “only 29% of applications are typically connected.” Those are large enterprises, but the pattern is familiar at any size.
TSIA's 2026 report says “the ability to unify customer data is one of the strongest predictors of renewal performance.” It publishes no figure for that. It also warns that “without data literacy, predictive alerts are ignored, and automation stalls.” Both point at the same thing. The dashboard is only as good as the records behind it, and the team's habit of acting on them.
How to rebuild a customer success dashboard
- List every tile. Run each through the five questions.
- Cut tiles that fail question 2. If nobody knows what to do when it moves, it's decoration.
- Add a target and an owner to every tile that's left.
- Put it where the work happens. A list of accounts that crossed a threshold, in the tool the team already uses, beats a screen on the wall.
- Review it in the same meeting every week, with the owner saying what they did.
You'll end up with fewer tiles. Our post on customer success metrics that matter covers which metrics are worth a tile, and our guide to customer operations KPIs covers support and onboarding measures.
A dashboard built on the work itself
The 24-tile dashboard pulled from six systems. Each refresh took an export. Nobody could click a tile and see the accounts behind it, so nobody did anything with it.
MatrixFlows keeps customer records, requests, onboarding projects and knowledge in linked tables in Matrix. Views and reports read from those same records, so a tile can open the list of accounts behind it. Automations can turn a threshold into a task for a named owner, with a due date. The action plan sits on the record, not in a separate slide.
Back to Carmen's question. The rebuilt dashboard had six tiles. Each had a target, an owner and an action. The Monday meeting started with them again.
Create a Free Workspace → and rebuild one dashboard tile with a target, an owner and an action this week.
.svg)
.svg)
.svg)
