Decision intelligence is the layer most business intelligence investment never reaches. Most large enterprises have spent two decades and tens of millions building business intelligence. The visible result is a sprawling estate of dashboards — often several hundred of them — refreshed every morning, watched by almost no one. And yet the important decisions are still made the same way they were before any of it existed: in a meeting, days after the data was available, largely on instinct. The dashboards multiplied. The decisions did not get faster, and they did not get better. That gap — between the analytics a company owns and the decisions it actually makes — is the problem the next stage of BI exists to close.
“By 2023, more than a third of large organisations will have analysts practising decision intelligence, including decision modelling.”
The Dashboard Was Never the Point
Nobody ever made a decision by staring at a dashboard. They looked at the dashboard, formed an impression, walked into a meeting, argued about what it meant, and decided something afterwards — often on the judgement of whoever spoke most confidently. The dashboard was an input to a human process that the dashboard did not change.
For a long time this looked like progress, because each new dashboard felt like more visibility. But visibility is not the constraint. Forrester estimates that between 60% and 73% of all data inside an enterprise is never used for analytics at all (Forrester). The bottleneck was never the supply of charts. It is the last mile — the distance between an insight existing somewhere and a decision actually changing because of it. Adding a four-hundredth dashboard does not shorten that mile. It usually lengthens it, because now the signal is buried in more noise, and someone still has to notice it, interpret it, and act.
A dashboard answers what happened. The business needs to know what to do. Those are not the same question, and no amount of additional dashboards turns the first into the second.
The Five Stages of the Evolution
Business intelligence has been climbing the same ladder for thirty years, from describing the past toward driving the decision. There are five recognisable stages.
- Reports. What happened. Static, periodic, backward-looking — the monthly pack. Useful, and entirely in the rear-view mirror.
- Dashboards. What is happening. Interactive and near-real-time, but still descriptive and still dependent on a human to notice the right thing and act. This is where the overwhelming majority of enterprises live today.
- Predictive analytics. What will happen. Forecasts, propensity scores, demand signals — the first stage that faces forwards instead of backwards.
- Recommendations. What you should do. Prescriptive analytics and next-best-action: the system does not just predict churn, it proposes the retention offer, with the expected impact attached. The human decides; the system has done the thinking.
- Autonomous decision support. The decision is made and executed within defined guardrails, and a human supervises by exception. Not “the machine takes over” — rather, the routine, high-frequency decisions run themselves, and people spend their judgement where it actually matters.
The decisive break is between stage 2 and stage 3. Everything up to dashboards describes the past and leaves the decision entirely to a human. Everything from predictive onward starts doing part of the decision work. Stages 3 to 5 are what the discipline of decision intelligence is about — and most companies have a vast investment parked at stage 2, mistaking more dashboards for movement up the ladder.
Why Dashboards Are Reaching Their Limits
Dashboards are not failing because they are badly built. They are reaching the limit of what a descriptive tool can do, for four structural reasons.
They look backward; decisions face forward. A dashboard shows what already happened. Every decision is a bet on what happens next. The translation from one to the other still lives entirely in someone’s head.
They put all the cognitive load on a human. A dashboard assumes a person will watch it, spot the thing that matters among everything that does not, and know what to do about it. At one dashboard, that is reasonable. At four hundred, it is impossible — and the important signal is statistically certain to be on a tab nobody opened today.
They are disconnected from action. The insight lives in a BI tool. The decision happens in a meeting, an ERP, a CRM, a pricing system — somewhere else, later. The handoff is manual, and most insights die in that gap.
They answer “what,” never “so what” or “now what.” A dashboard reports a 4% margin decline. It does not tell you which three actions would arrest it, what each would cost, or which to take first. That is the part the business actually needs, and it is the part a dashboard structurally cannot provide.
None of this means dashboards disappear. It means they stop being the destination and become one rung on a ladder most companies have stopped climbing.
What Decision Intelligence Looks Like in Practice
Three patterns show the move up the ladder — from describing a problem to driving the decision.
Pricing: from a margin dashboard to a recommendation. A retailer had an excellent margin dashboard that nobody acted on quickly enough, because turning it into price changes took a weekly committee. They moved the decision to stage 4: a system that proposes specific price adjustments with the expected margin and volume impact attached, which a category manager approves or overrides in minutes. The data did not change. The decision moved from a meeting to a workflow.
Replenishment: from stockout reports to autonomous reordering. A distributor watched stockouts on a dashboard and reacted after the fact. They moved high-frequency, low-risk replenishment decisions to stage 5 — automatic reordering within guardrails (value ceilings, supplier rules), with planners supervising by exception. The planners did not lose their jobs; they stopped spending them on the 90% of orders that never needed a human, and concentrated on the 10% that did.
Churn: from a churn dashboard to next-best-action. A telco had a churn-risk dashboard that the retention team rarely opened in time. They embedded stage-4 recommendations directly into the agent’s screen in the CRM — at the moment of contact, the system surfaced the specific retention offer most likely to work for that customer. The insight stopped living in a separate tool and started arriving where the decision was actually made.
In each case the unlock was not a better chart. It was moving the decision closer to the data and the action.
How to Start
You do not climb this ladder by buying a new platform or building more dashboards. You climb it one decision at a time.
- Start from a decision, not a dashboard. Pick one recurring, high-value decision — a pricing call, a replenishment order, a credit approval, a retention offer. Ask what that decision actually needs to be made well, and how long it currently takes. That decision, not your BI estate, is the unit of work.
- Move one decision up one rung. Do not try to make the whole organisation autonomous. Take a single decision from dashboard to recommendation: have the system propose the action, with expected impact, and let a human approve it. Prove the loop before you scale it.
- Put the recommendation where the decision happens. Insight in a BI tab gets ignored. The same recommendation inside the pricing system, the CRM, or the planning tool — at the moment of the decision — gets used. Move the intelligence to the workflow, not the workflow to the intelligence.
The higher rungs depend on foundations the dashboard era never required: governed, connected data, and the operational discipline to run a system that acts. The companies that get there treat BI not as a wall of dashboards to admire, but as a ladder of decisions to climb.
A useful diagnostic to leave the next analytics review with: count your dashboards, then count the decisions that actually changed last quarter because of one. If the second number is far smaller than the first, you do not have a business intelligence problem. You have a decision problem — and more dashboards will not solve it.
Ready to Assess Your Data and AI Opportunity?
The IQ Mart Data and AI Opportunity Scan identifies the highest-value decisions still trapped on a dashboard, and outlines a practical first step toward decision intelligence — in 30 minutes, without a sales pitch.