Build a dashboard your team actually opens

What you will have at the end.

A one-screen dashboard with four indicators, refreshed automatically, plus a one-page monthly report template you can fill in twenty minutes.

A dashboard is not a display of everything you can measure. It is a small set of numbers that, when one of them moves, someone does something different the next morning. If nothing would change, the number does not belong on it.

Choosing the four

1. Apply the so-what test to every candidate

   Take every number you are tempted to show and ask one question: if this went up thirty percent, what would we do differently. If the answer is nothing, or we would be pleased, it is not an indicator. It is trivia.

   Sessions, impressions, followers, page views and click-through rate all fail this test at a clinic. They are diagnostic numbers you look at when something is wrong, not numbers you watch weekly.

2. The four that survive for most clinics

   Different clinics land differently, and most land close to this set.

   —New patients booked, this month, from your practice software. The only number that is unambiguously true.

   —Cost per new patient, by service category. Spend divided by new patients from that category.

   —Enquiries not yet called back. A live number, and the only one on the dashboard that is about today.

   —Fill rate on the schedule, or the days to the next available appointment. Because spending money into a full schedule is a way of losing money slowly.

   In practice

   That fourth one is the indicator almost nobody puts on a marketing dashboard and the one that most often changes a decision.

3. Write the definition of each one down

   Every number needs a one-line definition, visible on the dashboard itself. Six months from now somebody will ask whether a new patient includes a returning patient after two years, and if the answer is not written on the dashboard the whole thing loses credibility in one conversation.

   —New patient: a person with no completed visit in the previous twenty-four months.

   —Cost per new patient: all category spend in the month, divided by new patients attributed to that category, using last click.

   —Say which attribution rule you used. Not because it is the right one, but so the number means the same thing each month.

Joining the sources

1. Accept that three systems will never agree

   Your practice software, your analytics and your ad platforms each count differently, over different windows, with different rules. They will never match and chasing a match is a way to spend a month.

   Pick one as the truth for each number and say so on the dashboard.

   —Patient counts: the practice software. Always.

   —Spend: the ad platform. Always.

   —Site behavior: analytics. Always.

   —Never divide a number from one system by a number from another without noting it. That is where most nonsense cost-per-patient figures come from.

2. Build it in a spreadsheet before you build it in anything else

   A spreadsheet with one tab per source and one summary tab covers what a clinic needs, costs nothing, breaks rarely, and can be understood by whoever inherits it.

   Move to a dedicated dashboard tool when the spreadsheet genuinely stops coping, which for most clinics is never.

   —One tab per source, pasted or pulled, raw and untouched.

   —One summary tab that reads from them. Never edit numbers in the raw tabs.

   —A date stamp showing when each source was last refreshed. A stale dashboard that looks live is worse than no dashboard.

3. Automate the refresh, or accept it will rot

   A dashboard that needs someone to update it manually every Monday will be current for three weeks. Either automate the pull or reduce the dashboard to what you can genuinely refresh.

   —Scheduled exports from the ad platforms into the raw tabs.

   —An analytics connection into its own tab.

   —The practice software export is often the one that stays manual. Make it a fifteen minute monthly job on a named person's calendar, not a good intention.

   —If a scheduled pull fails, somebody should be told. A silent failure produces a dashboard confidently showing last month.

The monthly report

1. Write it flat: the number, then the reason

   A monthly report is not a presentation. Lead with the figure, then say why it moved, then say what you are doing about it. No executive summary, no build-up, no chart that decorates a sentence.

   —New patients: 41, against 36 last month. Restorative campaign, which we raised on the 8th.

   —Cost per new patient: 118 dollars, up from 94. The rise is the new category, which is still in its first month.

   —Not: this month showed encouraging momentum across several key performance indicators.

2. Separate what you measured from what you think

   Three labels, used consistently, and the report becomes trustworthy in a way most reports are not.

   —Measured: you have the number and you can show where it came from.

   —Hypothesis: a plausible explanation you have not confirmed. Say so.

   —Undecidable: something you cannot know with the measurement you have. Say that too, and say what it would take.

   In practice

   The third label is the one that earns credibility. A report that admits it cannot attribute phone bookings is more believable than one that quietly attributes them anyway.

3. End with actions, not observations

   Three to five lines. Each one names what is being done, by whom, and by when. Anything without those three is not an action, it is a thought.

   —Raise the restorative daily budget to 42 dollars. Me. Monday.

   —Rewrite the hygiene landing page headline to match the ad. Provider. Before the 15th.

   —Pause the pediatric campaign, the schedule is out to five weeks. Me. Today.

When it goes sideways