Choose a marketing dashboard that leads to a decision

A useful marketing dashboard shows which question needs attention, where the data comes from and what you can do with it. Choose decisions and measurement definitions first, then select a reporting tool. For a small business, a clear worksheet can be more valuable than a screen with twenty charts.

The most attractive dashboard can conceal an important error: two columns use the same word but count different things. The advertising service calls a click a result, the form counts a submission, and the owner means a suitable enquiry. A good overview starts with that conversation.

Write down three decisions it needs to support

Begin with the work you want to assess over the coming period. Do you want to improve a landing page, adjust a campaign budget or choose newsletter content? For each question, record who decides and when they need the information.

A weekly overview is not useful for every decision. A technical fault may need immediate investigation. A service with a long sales process requires a longer view of enquiries and follow-up. The rhythm follows the decision, rather than the chart's refresh frequency.

Connect those questions to your marketing plan. If you do not yet know which problem marketing should help solve, every figure gets roughly equal attention. A dashboard should help limit priorities.

Make a small glossary before the first chart

For each key metric, establish meaning, source, period and owner. Record what is excluded too. An enquiry might mean a received message about an offered service, with spam and supplier emails recorded separately.

A measurement definition need not become a technical document. “Received contact forms, checked for a suitable service enquiry, per calendar month” says more than a column labelled only “leads”. Keep the definition somewhere report users can open it.

Spell percentages out fully. Ten enquiries from two hundred measured visits is five per cent. Ten enquiries from one hundred advertisement clicks is ten per cent. Both can be mathematically correct, but they answer different questions. Show counts and denominator together.

Revenue also needs agreement: do you mean quotations, orders, invoiced amounts or received payments? Keep these stages separate. The guide to measuring marketing results without false certainty helps define these terms.

Follow the path from attention to a suitable enquiry

Choose a limited number of signs that help identify a possible blockage. Reach may indicate visibility, visits an observed arrival, and enquiries a subsequent step. One number does not replace another.

Likes and impressions do not have to disappear. They are useful when your question concerns distribution or interest. They become misleading when used to prove campaign profitability without the additional data required.

Put a possible action beside each chosen measure. If you can never name a follow-up question or action for a number, it may not belong on the first page. Background information can remain available without constantly taking priority.

A fictional scorecard that reveals a problem

Voorbeeldstudio is an invented service business. All figures below are practice material, rather than a measured campaign or results promise. The business wants to know whether visitors find enough information to make a suitable enquiry.

The scorecard compares two equally long practice periods:

• Measured service-page visits: 200 in period A, 240 in period B. First follow-up question: were channel mix and measurement settings comparable?
• Received forms: 12 in period A, 8 in period B. First follow-up question: do the form and delivery still work as expected?
• Suitable enquiries under the agreed definition: 6 in period A and 6 in period B. First follow-up question: why has the proportion within all forms changed?
• Recurring question in suitable conversations: the service area in both periods. First follow-up question: is the area clearly stated on the page?

A visits-only chart would suggest progress. The scorecard gives another reason to investigate: suitable enquiries remain unchanged and the same practical question recurs. That makes the service-area explanation a possible improvement.

The example proves no cause. Voorbeeldstudio first needs to check measurement and the form, then examine conversations. The decision might be to clarify the service area on the page and review the questions again next time. That is more concrete than “achieve more conversion”.

Give data quality as much room as the figures

Record when tracking changed, a form failed or a source was temporarily missing. Without context, a measurement fault looks like a sudden change in interest. A short chart note can prevent a wrong decision.

Do not quietly compare different sources either. A platform may record clicks, while your website records only visits your setup can observe. Time zones, filters, consent and automated systems can affect counts. A discrepancy need not be an error, but requires explanation.

Use a consistent campaign naming agreement. The guide to using UTM tags shows how to keep source, medium and campaign recognisable. A dashboard can merge untidy labels, but a sound agreement at the entrance is easier to maintain.

Account for delays and duplicate attribution

An enquiry can follow weeks after a first contact. A newsletter, search and conversation may all have contributed. A channel report using its own attribution model tells you which share that model assigns.

Record the period and model you use. Do not add different platforms' attributed results together as though they were independent customers. Where possible, show an actual received enquiry once in your own overview, allowing for unknown origin.

With small numbers, a longer window sometimes helps more than another decimal place. One extra enquiry can create a large percentage rise. Include absolute counts and postpone major budget shifts when the evidence is still thin.

Choose the simplest tool that can handle the work

A worksheet suits a small number of sources, definitions still being refined and someone who can reliably maintain entries. A connected reporting tool may help when the same established data regularly needs gathering from several systems.

Assess maintenance effort. Who repairs a broken connection? Can you inspect the calculation used? Who has access, and which data is passed on? A cheap licence can still demand substantial follow-up.

Test a candidate using your own small scorecard. Ask someone else to find a number's source, period and definition. Check whether empty or missing values remain recognisable rather than silently appearing as zero. That says more about usefulness than the number of chart types available.

End every review with a decision and an owner

At an agreed time, return to the original question. First check whether the data is usable. Then describe what you see, which explanation you want to investigate and which action follows.

Give every action an owner, date and review point. “We will adjust the service-area explanation this week and review recurring questions next month” is actionable. An upward arrow beside a visitor count is not yet an action.

Before the meeting, use this short check:

1. Are periods, sources and definitions clear?
2. Are measurement faults and missing data noted?
3. Are counts shown beside percentages?
4. Is channel attribution recognisable as a model?
5. Can each key metric lead to an appropriate follow-up question?

Check technical problems through the actual visitor journey, using the post-publication check, for example. Find other measurement and interpretation questions under Measure and improve.

Sources and further reading

• Google Analytics: traffic sources and their different scopes.
• Google Analytics: attribution and assigning credit to touchpoints.