The Problem Isn’t the Tool — It’s the Dashboard Itself
According to Gartner, 70% of decision-makers say they don’t have access to the KPIs they need in real time. And yet, most companies have dashboards. Often too many.
The real problem: these dashboards are built backwards. They start with available metrics instead of business questions. The result? Screens full of numbers nobody looks at, weekly manual exports nobody reads, and hours lost maintaining outdated reports.
After designing and auditing over 150 dashboards for companies of all sizes, here are the failure patterns I find consistently — and the method for building a dashboard that actually drives decisions.
5 Signs Your Dashboard Is Useless
Sign 1: Nobody Checks It Regularly
If your dashboard is only opened during monthly meetings, it’s not for steering — it’s for justification. A useful dashboard is consulted daily or weekly by at least one person who makes decisions with it.
The test: check your dashboard’s access history. If the last visits were more than 7 days ago, you have a problem.
Sign 2: It Only Shows Vanity Metrics
Page views, follower counts, impressions… these metrics are reassuring but don’t guide any action. If your dashboard doesn’t contain at least 3 actionable metrics (conversion rate, CAC, LTV, ROAS, retention rate), it’s decorative.
The test: for each metric on the dashboard, ask yourself: “If this number drops 20%, what specific action do I take?” If you don’t have an answer, the metric doesn’t belong there.
Sign 3: Data Is Refreshed Manually
If someone spends 2-3 hours every week exporting CSVs, pasting them into Google Sheets, and updating charts, your dashboard is a productivity sinkhole. On average, marketing teams waste 12 hours per month on manual reporting.
The test: does your dashboard display a last-updated date? If it’s not “today” (or at most “yesterday”), the data is already stale.
Sign 4: It Mixes Too Many Levels of Detail
A dashboard with 40 widgets, 15 filters, and 8 tabs isn’t comprehensive — it’s unusable. The human brain can effectively process only 5-9 visual elements at once (Miller’s Law).
The test: show your dashboard to someone who’s never seen it. If they can’t understand the main message in under 10 seconds, it’s too complex.
Sign 5: It Doesn’t Answer Any Specific Question
The classic symptom: the dashboard displays data “just in case” rather than answering specific questions. “Here’s our data” is not the same as “Here’s whether we’re ahead or behind on our goals.”
The test: can you state in one sentence the business question each dashboard view answers? If not, it’s a report, not a decision tool.
How to Build a Dashboard That Works
Step 1: Start With Questions, Not Data
Before opening Looker Studio or Power BI, gather the decision-makers and ask:
- What decisions do you make every week? (increase/decrease a budget, launch/stop a campaign, prioritize a channel…)
- What information are you missing to make those decisions faster?
- What’s the primary business objective this quarter?
These answers determine 80% of your dashboard’s content.
Concrete example: an e-commerce business focused on profitability needs to see ROAS by channel, CAC vs LTV, and margin by category at a glance — not page views by source.
Step 2: Limit to 5-10 KPIs Maximum
The golden rule: if everything is important, nothing is. An effective dashboard contains:
- 3-4 strategic KPIs at the top (revenue, overall ROAS, conversion rate, margin)
- 5-6 operational KPIs below (by channel, category, segment)
- Zero metrics that don’t trigger action
| Type | Examples | Check Frequency |
|---|---|---|
| Strategic | Revenue, ROAS, Net margin, LTV | Daily |
| Operational | CPA by channel, conversion rate by device, average order value | Weekly |
| Diagnostic | Bounce rate by page, tracking errors | Monthly |
Step 3: Automate Data Refresh
A dashboard that isn’t current is a useless dashboard. Here’s the recommended architecture:
Data sources
↓ (automated connectors)
Data warehouse (BigQuery, Snowflake)
↓ (automated transformations)
Clean data layer
↓ (direct connection)
Dashboard (Looker Studio, Power BI)Practical rules:
- Strategic KPIs should refresh at minimum every 4 hours
- Use BigQuery as an intermediate layer to centralize sources
- Automate alerts when a KPI crosses a threshold (email, Slack)
- Eliminate all manual CSV exports from the process
Step 4: Design for Decision, Not Decoration
Design in service of action:
- Conditional colors: green/red based on defined thresholds (not arbitrary colors)
- Systematic comparisons: every KPI shown with prior period comparison (MoM, YoY)
- Sparklines: to see trends at a glance without taking up space
- Visual hierarchy: the most important KPIs are the largest and at the top
What to avoid:
- Pie charts — humans are bad at comparing angles
- 3D charts — always misleading
- Tables with more than 10 rows without a filter
- Y-axes that don’t start at 0 (unless justified)
Step 5: Structure in Depth Levels
A good dashboard is structured in 3 levels:
Level 1 — Executive view (1 page):
- The 3-4 strategic KPIs
- Readable in 10 seconds
- Answers: “Are things going well or poorly?”
Level 2 — Operational view (2-3 pages):
- Breakdown by channel, segment, product
- Answers: “Where’s the problem?” or “Where’s the performance coming from?”
Level 3 — Diagnostic view (link to analytics tool):
- Detailed explorations in GA4, SQL, or a dedicated tool
- Answers: “Why this problem?” and “How to fix it?”
Before & After: A Real Case Study
Before the Audit
A fashion e-commerce brand with $16M in revenue showed us their “main” dashboard:
- 47 widgets spread across 6 tabs
- Data refreshed manually every Monday (3 hours of work)
- Primary metrics: page views, sessions, bounce rate
- No view of ROAS by channel
- No comparison against objectives
Result: the dashboard was only viewed during the monthly committee meeting. Budget allocation decisions were made “by gut feeling.”
After the Redesign
- 12 KPIs across 2 views (executive + operational)
- Data refreshed automatically every 4 hours via BigQuery
- Primary metrics: revenue, ROAS by channel, CAC, margin by category, conversion rate by device
- Automatic Slack alerts if a channel’s ROAS drops below the profitability threshold
- Systematic year-over-year comparison
Result: the dashboard is checked daily by 4 people. The team identified within 48 hours a channel with negative ROAS that was burning $8,500/month. Weekly reporting time: from 3 hours to zero.
The Numbers That Sum Up the Stakes
- 70% of decision-makers lack real-time access to their KPIs (Gartner)
- 12h/month wasted on average on manual reporting (McKinsey)
- 23% improvement in decision-making with real-time dashboards (Forrester)
- 5x more likely to make better decisions with self-service data access (Harvard Business Review)
Where to Start
- List the 3 most frequent decisions your team makes
- Identify the 5 KPIs needed to make those decisions
- Audit your current dashboards: do they meet these needs?
- Automate data refresh
- Remove everything that isn’t actionable
Your current dashboards aren’t serving your decisions? We design custom, automated, action-oriented dashboards — connected to all your data sources.