Ask a room if they want real-time data and every hand goes up. Ask which decision changes in the next minute and they come down. Match freshness to the decision.

You do not need real-time data. You need right-time.

Ask a room of managers whether they want real-time data and every hand goes up. Ask which decision they would make differently in the next minute, that they could not make with this morning's numbers, and the hands come down. Real-time is the most requested and least interrogated thing in analytics. It sounds modern and ambitious, and most of the time it is neither needed nor used.

We have written more on this via our data platform and warehouse work, and Stop Paying For Fabric Capacity You Do Not Use takes a closer look at a related part of the picture.

It is also expensive. Real-time means always-on infrastructure, more complexity, more to monitor, more to break, paid for continuously whether or not anyone acts on it. It is a cost you take on in exchange for acting faster, and it only pays when there is a faster action to take.

The question that dissolves most of it

One question settles the majority of real-time requests. What will you do differently in the next minute, that you could not do with data from this morning? If the honest answer is nothing, and for strategic and management decisions it almost always is, then real-time buys you nothing but a bigger bill. A pricing review happens weekly. A board looks monthly. A sales manager acts on the day. None of those needs the number to change before their eyes.

The sensible frame is not real-time versus batch. It is right-time: the freshness each decision genuinely needs, matched to how fast you can actually act on it. Freshness should track the action window, not the ambition.

When real-time genuinely earns its place

There are real cases, and they are worth doing properly. Fraud and anomaly detection. Operational control of machinery or logistics. Customer-facing live status. Anything touching safety. The common thread is that a machine, not a person, acts on the data the instant it arrives. That is the tell. Genuine real-time drives an automated response, an alert, a block, a reroute, within seconds. If the plan is for a human to notice something on a screen, it is almost never real-time you need.

The dashboard nobody watches

The classic waste is a real-time dashboard built at considerable cost for someone to monitor, which nobody monitors, because nobody has a job that consists of staring at numbers as they tick. The value in time-sensitive data is almost always in automated action, an alert that fires, a threshold that triggers, not in a human watching. If a real-time requirement cannot name the automated action it drives, it is a screen, not a system, and it will be ignored within a week.

And the money is not free. Real-time competes, for budget and attention, with getting the fundamentals right: a clean model, trusted definitions, reports people believe. Spent on a live dashboard nobody watches, it is money taken straight from the work that would have mattered.

What to ask the next time someone requests it

So the next time someone asks for a real-time dashboard, do not start costing infrastructure. Ask them the one question: what would you do in the next minute that you cannot do now. If they can name an automated action, build it properly. If they cannot, you have just saved a permanent tax on a screen nobody would have watched, and you can spend it on the model underneath instead. Matching freshness to the decision is part of how we scope every data platform build and every Analytics Acceleration Programme, and the fuller version is in our Real-Time or Right-Time guide.

If any of this sounds familiar, talk to us about your data.

Related reading

• You Probably Do Not Need Microsoft Fabric Yet
• Per-User Or Capacity? The One Threshold That Decides
• Getting your Business Central data into Power BI: the options