Somewhere in your company there is a dashboard. It took weeks to build. It has tabs, filters, and a logo in the corner. Almost nobody opens it.
This is not a tooling problem. Looker, Tableau, Power BI, a spreadsheet with charts, the platform does not matter. Dashboards go unread for the same handful of reasons everywhere.
Here are the five we see most, and what to build instead.
The five failure modes
1. Nobody owns it
A dashboard without an owner is a poster in a hallway. People walk past it. Nobody is responsible for what it says or what happens next.
Ownership means one named person who reads the numbers first, explains what changed, and assigns the follow-up. When the answer to "who owns this dashboard" is "the team," the real answer is nobody.
The test is simple. If the dashboard broke tomorrow, who would notice first? If you cannot name that person, you have found the problem.
2. It has no narrative
A wall of charts is data. A report is an argument. Here is what happened, here is why, here is what we are doing about it.
Most dashboards skip the argument. They hand the reader twelve charts and expect them to do the analysis themselves, live, in a meeting. Nobody does that work. So people nod at the screen and move on.
The number is never the point. The sentence about the number is the point.
3. The data is stale, so nobody trusts it
Trust dies fast. Miss one refresh, or show one number that contradicts what the ad platform says, and the reader quietly stops believing the whole page.
Stale data almost always traces back to manual steps. Someone has to export, paste, and update, and that someone got busy. Every manual step in a reporting pipeline is a future gap in the data.
Once trust is gone, people route around the dashboard. They pull their own numbers, which are slightly different, and now every meeting opens with a debate about whose numbers are right instead of what to do next.
4. Metric soup
Forty metrics on one screen means no metric matters. Impressions next to cost per acquisition next to bounce rate next to follower count, all sized the same, all shouting.
Metric soup is usually diplomacy. Every stakeholder asked for one more number, and nobody was allowed to say no. The dashboard became an archive of requests instead of a tool for decisions.
A useful report makes each metric earn its spot by answering one question. Would a change in this number change what we do next week? If not, cut it or push it to an appendix.
5. It was built for the builder
Analysts build dashboards the way analysts think. Filters everywhere, every dimension exposed, maximum flexibility. That is a workbench, not a report.
The reader does not want flexibility. The reader wants the answer. Every filter you make the reader operate is analysis you shifted onto the person least equipped to do it.
Build the workbench for the analyst. Build something different for everyone else.
What a read report looks like instead
The fix is not a prettier dashboard. It is a different artifact with different rules. We treat reporting as part of the operational layer under marketing, the same layer covered on our marketing systems page, because that is where it lives.
It leads with sentences, not charts
The first thing the reader sees is written. Three to five plain statements about what changed, why it changed, and what happens next. Charts support the sentences. They do not replace them.
This is where AI earns its keep inside the engine. A system can draft the first pass of that narrative from the data every week, and the owner edits it in minutes instead of writing it from scratch. Reporting output multiplies without adding head count. We build exactly this in our analytics and reporting work.
It has an author and a deadline
A read report ships on a schedule, from a person. Monday at 9, from the owner, every week, no exceptions.
The deadline does the forcing. Because the report must ship, the pipeline has to be automated. Because it has a name on it, the narrative has to hold up.
It is scoped to decisions
Five to nine numbers, chosen because a move in any of them triggers a known action. Spend pacing, cost per lead, conversion rate, pipeline created. The right list depends on the business, and it should be short enough to argue about.
Everything else lives one click deeper for the people who need it. The workbench still exists. It is just not the front page.
It updates itself
No exports. No pasting. Data flows from the platforms into the report without a human in the loop, and alerts flag anomalies before the reader stumbles into them.
When the pipe is automated, stale data stops being a weekly incident and becomes a rare bug with a clear owner. This is the difference between reporting you maintain and reporting that runs.
A quick test for your reporting
Run three checks this week. They take an hour.
- Check the open rate. Most BI tools show who viewed what. Pull the last thirty days and count actual readers, not licensed seats.
- Ask three people the headline. Ask what the most important number was last month and what it did. If you get three different answers, there is no narrative.
- Find the last decision it caused. Look for a budget shift, a paused campaign, or a landing page change that cites the report. If you cannot find one, the report is decoration.
If the checks come back ugly, the problem is not your marketing team and it is not the tool. It is the layer underneath, the one that decides how data moves, who reads it, and what happens after. That layer is fixable, and fixing it is the whole job of our marketing operations audit.
Where to start
Unread dashboards are rarely the only symptom. The same broken layer usually shows up in messy naming conventions, attribution nobody believes, and hours lost to manual updates.
Our Performance Audit looks at how your marketing actually runs, reporting included, and hands you a dollar-weighted plan ranked by what each fix is worth. You keep the plan whether or not we do the work.