Insights

A Dashboard Shows What Happened While an Intelligence Brief Explains What Changed

Dashboards preserve visible state. Intelligence briefs isolate a material change, test possible explanations against evidence, and state the decision that follows.

A conversion rate falls. The dashboard shows the date, segment, channel, and size of the movement. The next question is why.

That question cannot be answered by making the dashboard denser. It needs a different object: an intelligence brief that isolates the change, checks the measurement, reviews relevant customer and company evidence, and states which decision is now exposed.

The dashboard and brief should stay connected. They should not be asked to do the same job.

The Dashboard Is the Observation Surface

A useful dashboard makes current state and movement visible without requiring a meeting. It keeps definitions stable, supports filters, shows history, and lets a reader inspect the underlying measure.

That consistency is a strength. The same chart should not quietly change its logic every time the company develops a new theory. If a reader filters by segment or date, the measure should remain comparable.

Google's Looker documentation describes dashboards as surfaces for viewing, filtering, drilling into, refreshing, and downloading data. Those are observation and exploration jobs. See the official Looker dashboard documentation.

The dashboard can show a break in the pattern. It may also carry an annotation or automated insight. The reviewed explanation still needs its own evidence and scope.

The Brief Begins With One Material Change

An intelligence brief should not summarize the entire dashboard. Choose the change that could alter a current decision.

State the measure, baseline, comparison period, affected segment, size of the movement, and first date it became visible. Preserve the exact dashboard link, filter values, data refresh time, and metric definition used during review.

Then state the decision at risk. A decline in qualified conversion may expose the campaign promise, audience definition, landing page, sales follow-up, or measurement setup. Naming the decision prevents the brief from becoming a list of interesting observations.

Check the Measurement Before Explaining the Market

A changed number can come from changed behavior, changed collection, or changed reporting logic. Run the data-quality check before asking the company to react.

Confirm the metric definition, source completeness, event instrumentation, attribution rule, denominator, time zone, refresh lag, and filter state. Check whether a release, consent change, CRM field update, or channel naming change altered what the dashboard receives.

Google Analytics documents that reports and explorations can differ because of supported fields, filters, date ranges, privacy treatment, modeling, and processing time. The lesson is broader than one platform: the reporting surface is part of the evidence. See the official reporting-surface comparison.

If the measurement changed, the brief may end there. Correct the data path and record the affected period instead of inventing a buyer story.

Treat Automated Explanations as Leads

Modern analytics tools can flag anomalies and identify dimensions associated with a change. That can shorten the investigation.

Microsoft Power BI, for example, can analyze an increase or decrease and rank the categories whose contribution changed most. Its documentation also states that the analysis is contextual to the prior data point and is unavailable in several scenarios. See the official Analyze feature guidance.

A ranked contributor is not automatically a cause. It tells the reviewer where to inspect. The brief should record the candidate explanation, the fields included in the analysis, known limitations, and evidence that supports or opposes the interpretation.

Add the Company Events That the Chart Cannot See

Many changes have an operational event nearby: a pricing update, product release, campaign launch, sales territory change, partner promotion, or site redesign. Preserve the event date and scope without assuming it caused the movement.

Google Analytics annotations exist for exactly this reason. They let teams place dated notes on reports to identify launches, explain spikes or dips, and record other events that may matter. See the official GA4 annotations documentation.

An annotation gives the chart memory. The intelligence brief checks whether the timing, affected audience, and mechanism actually fit the observed change.

Bring in Customer Evidence and Counterevidence

Performance data shows behavior at scale. Customer conversations can reveal the condition behind that behavior.

If conversion falls after a message change, inspect recent objections, proof requests, sales responses, support friction, and loss notes for the affected segment. Keep the exact passages and source dates attached. Count independent accounts separately from repeated records inside one account.

Look for evidence that weakens the leading explanation. The same objection may have existed before the metric moved. The affected campaign may serve only one portion of the declining segment. A product issue may fit the timing better than the page change.

The objection feed and customer language drift report provide source-linked records for this review. The brief selects only the evidence that changes the explanation.

Build the Explanation From Weighted Evidence

The explanation should show how much of the movement each known factor can account for and what remains unresolved. A waterfall view is useful because it preserves the starting measure, measured contributions, and current result without pretending every residual has a name.

Keep four statements separate:

Observed: the metric changed inside a defined scope.
Associated: a dimension or event moved with it.
Supported: independent evidence fits the proposed mechanism.
Verified: the evidence design can justify the stronger claim.

Most operating briefs will stop at supported. That is enough to make many reversible decisions. It is not permission to write a causal claim the evidence cannot carry.

State Competing Explanations Before Recommending Work

A brief becomes more reliable when it names the strongest alternative explanation. The purpose is not to create artificial balance. It is to show what the current evidence can distinguish.

For each explanation, record the expected pattern, supporting evidence, opposing evidence, and missing check. If two explanations remain plausible, choose a decision that produces information without creating an expensive commitment.

A narrow message test may be reasonable while a full positioning change is not. A verified tracking defect may justify an immediate correction. A thin customer pattern may justify another evidence window.

Trigger Briefs Selectively

Not every dashboard movement deserves a document. A brief is warranted when the change crosses a material threshold, persists beyond expected variation, affects a named decision, or conflicts with customer evidence.

Automated alerts can open the review. Google Analytics custom insights allow teams to define conditions that detect important changes, while automated insights identify unusual changes or emerging trends. See the official Analytics Insights documentation.

The trigger should carry the metric, scope, dashboard state, and threshold into the brief. A reviewer then confirms whether the movement is real and decision-relevant. Routine noise stays on the dashboard.

End With a Decision, Not an Explanation Alone

The brief should name the affected object, proposed action, owner, review state, expected result, and evidence that would reopen the decision. Include no action when the evidence supports keeping the current work.

Link the decision back to the exact dashboard state and every source record used. Later results return to the same brief without rewriting the original explanation.

Do not claim the action caused the later result unless the company ran a valid test. Record what changed, what remained stable, and whether the original explanation gained or lost support.

Keep the Two Objects Connected

The dashboard remains live. The brief remains reviewable.

Northstar's Glass Dashboard keeps measures, owners, work state, and historical movement visible. The Listening Engine supplies the source-linked customer and company evidence needed to investigate a meaningful change.

The dashboard tells the company where the record moved. The intelligence brief explains what the evidence says about that movement and which decision should move next.

Frequently asked questions

What is the difference between a dashboard and an intelligence brief?

A dashboard shows current and historical measures through a consistent reporting surface. An intelligence brief investigates one material change, states the evidence and limits behind its explanation, and names the decision that should follow.

Should an intelligence brief replace a marketing dashboard?

No. The dashboard remains the observation and exploration layer. The brief links back to the exact dashboard state, then adds the reviewed evidence and judgment needed for a specific decision.

When does a dashboard change need an intelligence brief?

Create a brief when the change is material, decision-relevant, not already explained by a known data issue, and likely to alter current marketing work. Routine movement can stay on the dashboard.

Can an automated insight explain why a metric changed?

It can identify unusual movement and associated dimensions worth investigating. Treat the output as a candidate explanation until data quality, timing, customer evidence, company events, and counterevidence have been reviewed.

What should an intelligence brief include?

Include the observed change, dashboard link and filter state, data-quality checks, relevant events, customer or company evidence, competing explanations, confidence language, affected decision, owner, and review condition.

How should a brief describe causality?

State what was observed separately from what is associated, inferred, or verified. Do not use causal language unless the evidence design can support it.

How does Northstar Stack connect dashboards and briefs?

The Glass Dashboard preserves current measures and work state. The Listening Engine supplies source-linked customer and company evidence, while the brief records the reviewed explanation, decision, owner, and returned result.