Start with the decision, not the dashboard

Most monthly reporting meetings contain one uncomfortable moment. The numbers have been presented, the charts have been explained, someone has apologised for a discrepancy between two systems, and then a director asks the only question that really matters: so what are we doing differently?

The room goes quiet.

That silence usually gets blamed on the data. It rarely deserves all the blame. In many mid-market businesses the reporting is imperfect but adequate. What is missing is any prior agreement about which decisions the reporting exists to support, who makes them, and what evidence would be enough to change someone's mind.

There is evidence that this gap between analytics and action is widespread. Gartner found that recommendations produced by marketing analytics are rejected by decision-makers 76% of the time. Its research points not simply to bad data, but to dashboards overwhelmed with information and insufficiently designed around the actions marketers actually need to take.

More data does not necessarily solve that. Supermetrics' 2025 Marketing Data Report, drawing on aggregated data from 6,000 businesses and a survey of more than 200 marketers, found marketers are using 230% more data than in 2020, yet 56% say they don't have enough time to analyse their data thoroughly.

So if your reporting feels busy but useless, work backwards. List the recurring decisions your business actually makes about growth, name the person who makes each one, then ask what result would change the answer.

Build reporting to serve those decisions.

Everything else is monitoring, and monitoring should be cheap.

The dashboard is a symptom, not the project

The usual response to unsatisfying reporting is to commission more reporting. A new dashboard. A BI licence. An analyst. A tracking rebuild.

Sometimes those are the right answers, and later in this article there is a test for when they are.

More often the same meeting happens again with better charts.

The reason is simple: a dashboard can only answer questions someone has asked. If the questions were never written down, the dashboard defaults to whatever the tools make easy to count, which is activity. Sessions, impressions, open rates, form fills, MQLs. All real, all measurable, none of them a decision.

Gartner's research into dashboard adoption reaches much the same conclusion: dashboards crowded with data can overwhelm marketers rather than drive action, particularly when they are designed around available data rather than stakeholder needs and decision flow.

This is not an argument against measurement. It is an argument about sequence.

Decisions first, then the evidence those decisions need, then the tooling.

Four different problems hiding behind "our reporting isn't good enough"

When a leadership team says reporting is the issue, they are usually describing one of four situations. They need different fixes, and confusing them is expensive.

"I don't trust the numbers"

Your analytics platform, your ad platforms and your CRM will disagree, because they count different events at different moments using different attribution windows and different definitions of a conversion. That is normal, not a fault.

But genuine data-quality problems are widespread too. Adverity's 2025 research among 200 CMOs across the US and Europe found that they estimated, on average, 45% of the marketing data used to make decisions was incomplete, inaccurate or out of date. Improving data quality was also the most commonly identified way of improving marketing performance.

The distinction matters.

The useful question is not "are our numbers perfect?" It is "is the problem big enough to change a decision?"

If two systems differ by fifteen per cent, and the decision only flips when a channel performs twice as well as another, then reconciling them is a distraction dressed up as rigour. Agree one system of record per decision rather than trying to make every system agree on every metric.

Where the gap does change the answer, fix it properly, and start with definitions rather than dashboards. Two teams counting "qualified lead" differently will produce two irreconcilable truths no matter how good the reporting layer is.

"Nobody looks at the dashboard"

Usually because it reports on things the viewer does not control.

A regional sales director cannot act on blended cost per lead. A performance marketer cannot act on quarterly revenue. If the person looking at a report has no lever attached to it, it is a status update, and status updates belong in a document, not a live dashboard.

This is where the 76% Gartner statistic becomes particularly interesting. The challenge isn't simply getting more analytics in front of decision-makers. It is producing evidence relevant enough to the decision that someone actually uses it.

"It takes three days to produce"

That is a real cost and worth removing.

But automating a report nobody acts on only makes the waste faster and harder to notice. Decide what the report is for before you industrialise it.

And the problem is increasingly unlikely to be a shortage of data. Supermetrics found marketers are now working with 230% more data than they were in 2020, while more than half say they lack sufficient time to analyse it properly.

The scarce resource is increasingly attention, not information.

"We can't tell what's actually working"

This is the one worth taking seriously, and it is rarely a shortage of data. It is usually a missing join.

The marketing activity is recorded in one place, the enquiry in another, and what sales did next in a third, with nothing reliable connecting them. Every team can describe its own performance. Nobody can describe the journey.

HubSpot's 2025 State of Marketing research found that only one in five marketers said their marketing data was fully integrated with the tools they use. Meanwhile, 30% described interactions between their marketing tools as disjointed, 28% struggled to get data they needed from other teams, and 34% struggled to share their own data across the organisation.

So before buying another analytics layer, find the missing join.

Build a decision inventory before you build anything else

Sit down with the people who actually make growth decisions and write the decisions out. Most businesses find ten to fifteen recurring ones. They tend to look like this:

• Do we keep funding this channel next quarter, hold it, or move the budget?
• Is the pipeline gap a volume problem or a conversion problem?
• Which segment or market gets the next campaign?
• Do we invest in sales support instead of more demand?
• Which content or sales asset do we retire?
• Do we hire a specialist or bring in a partner?

For each one, answer five questions:

1. Who owns this decision? A name, not a committee.
2. How often is it genuinely made? Not how often is it discussed.
3. What are the realistic options? If there is only one, it is not a decision.
4. What result would make us choose differently? Name the threshold.
5. What is the smallest piece of evidence that gets us there?

Question four is where reporting projects usually break open.

If nobody can say what number would change their mind, the decision is not being made on evidence, and no dashboard will change that.

That is not always a failure. Some decisions are properly matters of judgement, strategy or timing. It is far better to say so than to build a reporting layer that pretends otherwise and quietly gets ignored.

Question five tends to shrink the project. A decision about whether to keep funding a channel often needs one reliable join and a six-month view, not a real-time dashboard with forty tiles.

Match the reporting rhythm to the decision, not the calendar

Monthly reporting exists mostly because months exist. It is a poor fit for many growth decisions.

If a decision is genuinely made once a quarter, monthly reporting on it invites premature reaction to noise.

If your sales cycle runs nine months, this month's closed revenue is telling you about marketing you did last year, and judging current activity by it will make you cut the wrong things.

That is where leading indicators earn their place: not because early numbers are more accurate, but because they arrive in time to inform a decision that is actually open.

Ask of every recurring report:

Which decision does this feed, and does it arrive before that decision is made?

Anything that fails both parts of the question can move to a quarterly review or stop.

From "where might we be losing money?" to a decision someone owns

Crowdcube is a useful example of what this looks like when it goes well.

The team had marketing and sales activity running in parallel across a pan-European operation, with different audiences, entry points and sales processes, and a suspicion that money was being left on the table. What they lacked was visibility of how prospects were entering the funnel, what happened next, and where valuable opportunities were being lost.

The work that helped was not a bigger reporting layer.

It was mapping the key user journeys across markets, identifying the main entry points bringing people in, building segmentation around differing needs and potential value, and connecting marketing form submissions with the sales processes that followed, which showed where lead context and follow-up needed to work harder.

The output was a segmentation and content strategy the team owns and applies internally. The more valuable shift was in the conversation itself, which moved from "where might we be losing money?" to an evidence-led plan for what to change next.

The measurement mattered because it was built around a question the team was already trying to answer.

Design reporting around the person who has to act

Kefron shows the other half of the problem.

As the business expanded its services and markets, marketing, sales and customer teams were working in disconnected systems, with manual processes, spreadsheets and inconsistent data with no reliable single source of truth. Nobody had a complete view of the customer journey.

Rebuilding it on a single platform involved auditing and cleansing legacy CRM data, then designing pipelines, lifecycle stages and data structures capable of supporting multi-touch attribution.

The detail worth borrowing is the last part: role-specific dashboards.

Not one executive dashboard for everyone, but a view per role, built around what that person is expected to decide.

A dashboard designed to serve four audiences generally serves none of them. It becomes a compromise, so each group mentally filters it, and then rebuilds the bit they care about in a spreadsheet.

That spreadsheet is your real reporting system, and it is worth finding out who is maintaining it and why.

When better tracking really is the answer

There are clear cases where the reporting itself is the constraint and investment is justified:

• A specific decision is blocked by a missing join, most often between marketing activity, lead capture and what sales did next.
• You are about to make a large or hard-to-reverse investment, and the cost of being wrong exceeds the cost of measuring properly.
• Two teams use the same words for different objects. Definitions first, then systems.
• The lag between action and result is longer than your patience, so you need agreed leading indicators to avoid judging good activity too early.

And the cases where it probably is not:

• The decision has no owner. Fix that first, because it is free.
• Nobody can name a threshold that would change the answer.
• The systems disagree, but not by enough to change what you would do.
• What you actually want is confidence in a decision you have already made. That is a reasonable human need, but it is not a reporting requirement, and a project justified on those grounds will disappoint everyone.

You may not need a new dashboard.

You may need six named decisions, two agreed definitions and one connection between systems you already own.

A first pass you can complete in a fortnight

1. Write the decision list with the people who make those decisions, not on their behalf.
2. Attach every existing recurring report to a decision. Anything with nothing attached moves to quarterly or stops.
3. For the two or three decisions that matter most, write the threshold in a sentence a colleague would recognise.
4. Write down definitions for lead, qualified, opportunity and won, and get sales and marketing to sign the same sheet of paper.
5. Identify the single missing join that blocks the most decisions.
6. Only then decide whether you have a tooling problem.

Steps one to four cost nothing but attention, and they frequently remove the need for step six.

The question worth asking in your next reporting meeting

Before the next reporting cycle, take one number your team reviews every month and ask:

If this came back twice as good, or half as good, what would we do differently, and who would decide?

If the answer is clear, the reporting is doing its job.

If nobody can answer, you have found the real gap, and it is not in the data.

If your reporting produces plenty of numbers but few decisions, the useful next step is usually to look at how measurement, systems and commercial decision-making connect, rather than at the dashboard itself. That is the territory covered in how FutureGroup approaches making growth repeatable.