Ask a leadership team a basic operating question, what sold last week, in which channel, at what margin, to which kind of customer, and watch what happens. Someone opens the commerce platform. Someone else opens a spreadsheet. Finance has a third number that disagrees with both, and the meeting spends its first twenty minutes negotiating whose figure is real.
We have built intelligence hubs that end that meeting, and the work has settled into a repeatable shape. This article gives the shape away.
What is an operating picture? #
An operating picture is a single, live view where sales, stock, channels and customers appear as one connected story rather than four exports. One screen, current numbers, agreed definitions. Its test is behavioural: the leadership meeting stops reconciling and starts deciding, because the arguing-about-numbers phase has been engineered away.
The word "picture" is doing deliberate work. A pile of dashboards is an art gallery: each one accurate, none of them composed together. A picture has relationships. Sales joined to stock tells you what to reorder and what to discount. Channels joined to margin tells you where growth is actually profitable. Customers joined to all of it tells you who to keep and what they will want next.
Most businesses already own every number the picture needs. The numbers are just imprisoned in systems that do not speak, reconciled by hand once a month, and stale by the time anyone senior reads them. The picture is a plumbing achievement before it is an intelligence one.
What should a Monday meeting answer in ten minutes? #
A leadership meeting reading one operating picture should answer eight questions in ten minutes: what sold, at what margin, through which channels, against what stock position, from which customers, at what acquisition cost, with what change from last week, and what single number most needs a decision today.
Write those eight on a whiteboard and score your current Monday against them. Most teams find they can answer two or three quickly, another two with a day's lag and a spreadsheet, and the rest only quarterly, if ever. The gap between ten minutes and "we would need to pull that together" is the size of the opportunity.
The eight questions are also the correct specification document for the build. Tools change, vendors change, but a hub built to answer the leadership team's actual weekly questions cannot drift into becoming a vanity dashboard. The full list is below; steal it and run your next Monday against it before you build anything.
What is the right build order? #
Integration first, then insight, then operations. Integration connects the systems so the numbers agree and update themselves. Insight builds the reads on top: the picture, the questions, the alerts. Operations closes the loop: the business acting on the picture automatically, reorders, follow-ups, escalations. Each layer only works standing on the one beneath it.
Integration is the unglamorous majority of the work and the least skippable. It is where definitions get fought over ("what counts as a sale, and when?"), where the duplicate customer records surface, where the spreadsheet that secretly runs the business gets found and retired. Everything credible about the later layers is earned here.
Insight is where the leadership team feels the change: the Monday meeting shrinks, the picture replaces the reconciliation. Operations is where the compounding starts, because a picture that acts on itself gives back hours as well as clarity. This is the same Integration, Insight, Operations sequence we run inside AlchemAI, and it exists because we kept watching what happens when people skip ahead.
Why does skipping ahead always fail? #
Skipping ahead fails because each layer inherits the honesty of the one below. Insight built on unintegrated data produces a beautiful picture of wrong numbers, and the first time an executive catches it wrong, trust is gone for a year. Operations built on untrusted insight automates mistakes at machine speed.
The failure is seductive because the demo works. Dashboard tools will happily visualise disconnected exports, and for one meeting it looks like the future. Then the numbers drift from the source systems, someone reconciles by hand to check, the old spreadsheet comes back, and the dashboard joins the graveyard. The business concludes "we tried that", which poisons the well for the version that would have worked.
The discipline costs patience up front and repays it permanently: the boring connection work first, the picture second, the automation third. Businesses that hold the order end up reading themselves in ten minutes a week. Businesses that skip it end up with a subscription and a story.
A business that can read itself in ten minutes spends the rest of the week acting like it.