Short answer: the four types of analytics are a ladder, not a menu. Descriptive tells you what happened (this block yielded 7.2 tonnes per hectare). Diagnostic tells you why (the ferment stuck because YAN was low and the must ran warm). Predictive tells you what will happen (Block 4 hits 24 Brix in nine days; this lot is at elevated VA risk). Prescriptive tells you what to do (pick Block 4 first, run this blend, barrel the press fraction separately). Each rung stands on the one below — and Microsoft Fabric is interesting to a winery precisely because it puts the whole ladder on one platform instead of a vineyard spreadsheet, a cellar logbook and a sales system that never speak.
This post opens a three-part series on process intelligence for wineries — the gap between having data and having answers from the vineyard to the bottle. Part one is the map: the four classic types of analytics, translated out of consulting-speak into Brix, ferment and barrel language, with an honest look at where Microsoft Fabric’s newer capabilities fit. Part two asks why so many Power BI dashboards die in wine businesses, and part three is what changes when the developer has stood on the crush pad.
Descriptive: what happened
The foundation rung, and the one most wineries genuinely need first. Descriptive analytics is the faithful mirror: tonnes picked and yield per hectare by block and varietal; fermentation count and tank utilisation through the crush; barrel and tank stock by vintage, varietal and cooperage; losses from intake through racking to bottling. Nothing clever — and enormously valuable, because a winery’s data is usually scattered across a vineyard spreadsheet, a cellar whiteboard and a sales system that have never been in one room. This is the “see it” rung of where a winery starts with AI, and it is where Power BI lives most comfortably.
The oenology-trained instinct matters even here: which numbers you describe is a domain decision. Yield per hectare means little without the block’s vine age and spacing; a “tonnes” figure is fruit weight, not finished volume, and the conversion through press yield and ageing losses is where naive stock reports go wrong.
Diagnostic: why it happened
The rung where dashboards earn their keep. This ferment stuck at 4 Brix — why? Diagnostic analytics is the structured walk back through the lot’s history: was yeast-assimilable nitrogen (YAN) low at inoculation (a nutrient problem)? Did the must run hot and stress the yeast (temperature control)? Was the nutrient addition mistimed? The data requirement is unforgiving: you need the lot genealogy — this block, this pick, into this tank, with these additions and this temperature trace, racked to these barrels — joined end to end. Without that chain, “why did this lot go wrong” dies as an argument at the morning cellar meeting. With it, it becomes a five-minute drill-down across Brix curves, YAN, temperature and addition logs.
Predictive: what will happen
The rung machine learning lives on, covered across this blog: harvest date and grape ripeness from Brix, acidity and weather; vineyard yield months ahead; fault risk — Brett, VA, TCA before it becomes irreversible. Prediction needs history — measured Brix series, real lab panels, honest fault records — which is why it sits above descriptive and diagnostic, never instead of them. A model trained on a patchy vintage log is a mirror of bad data, confidently wrong, exactly as the machine-learning foundations post warns.
Prescriptive: what to do
The top rung and the most oversold. Prescriptive analytics turns a prediction into a recommendation: given ripeness curves across blocks and limited tank space, this is the pick order; given lot chemistry and the house style, these blend fractions score best in trial; given predicted drinking windows, barrel this lot in new French oak and that one in neutral. Done honestly, it’s optimisation over predictions you already trust, with the winemaker ranking the shortlist. Done dishonestly, it’s a black box telling a winemaker how to blend — and the correct response to that is the scepticism it will receive. Prescriptive output belongs as a ranked recommendation with its reasoning showing, owned by a human, especially anywhere near a blend decision or a fault call.
Where Fabric actually fits
Microsoft Fabric’s pitch to a winery is not any single feature — it’s that the four rungs stop being separate, disconnected systems. OneLake holds one copy of vineyard, cellar and sales data instead of a spreadsheet, a logbook and an ERP that never reconcile. Eventstreams and Real-Time Intelligence take live readings from fermentation probes and cold-room sensors — the IoT layer I covered in sensors in the winery — without a custom streaming stack. Notebooks train the predictive models on the same data the dashboards read, so the harvest forecast and the cellar report never disagree about what’s in a tank. Direct Lake lets Power BI read the lakehouse without nightly refresh gymnastics. And Data Activator turns a fermentation temperature breach or a rising VA reading into an alert to the cellar hand on shift — the difference between analytics and wallpaper. The full catalogue is in Microsoft Fabric for wineries: 20 use cases; the point here is the ladder view: one platform, four rungs, one copy of the truth.
Where this breaks
The honest section, as always. The ladder is climbed in order, and most attempts aren’t. A prescriptive blend-optimiser bought before the lab and addition history is clean is an expensive way to automate guesswork. Diagnostic depends on lot genealogy nobody glamorous wants to build — the block-to-barrel chain is unsexy integration work, and skipping it caps you at descriptive forever. Prediction inherits every flaw in the record — a vintage is one data point a year, so wine models are perpetually data-hungry and humble; ten vintages is a thin training set by any standard. And the vintage itself moves the target — a model learns the seasons it saw, and a genuinely unusual year is exactly when you most want a forecast and least should trust one. The platform doesn’t fix any of this; it just removes the excuse that the spreadsheet, the logbook and the ERP couldn’t talk.
The bottom line
Descriptive, diagnostic, predictive, prescriptive — what happened, why, what’s next, what to do. A winery that climbs in order turns its cellar meetings from arguments into drill-downs, its harvest from a scramble into a forecast, and its blending from instinct-only into instinct-plus-evidence. Fabric’s contribution is putting the whole ladder on one platform with one copy of the data. What it cannot do is supply the oenological judgement that decides which numbers matter and whether they can be trusted — and that gap is exactly where most dashboard projects in wine quietly die. That’s the next post.
Frequently asked questions
What are the four types of data analytics for a winery? Descriptive (what happened — tonnes picked, yield per hectare, tank and barrel stock), diagnostic (why it happened — tracing a stuck fermentation or a faulty lot back through Brix, YAN, temperature and additions), predictive (what will happen — harvest date from ripeness curves, fault risk, when a wine is ready) and prescriptive (what to do — pick order, blend composition, barrel allocation). Each builds on the one before it.
How does a winery use the four types of analytics? Descriptive: yield per hectare by block, fermentation count by tank, stock by vintage and varietal. Diagnostic: why this ferment stuck — low YAN, temperature spike, nutrient timing. Predictive: harvest date from Brix and weather, Brett or VA risk, drinking window. Prescriptive: pick order across blocks, blend trials ranked, which lots to barrel where. The order matters — you cannot prescribe from data you cannot yet describe.
Where does Microsoft Fabric fit in winery analytics? Fabric puts the whole ladder on one platform: OneLake as the single copy of vineyard, cellar and sales data, eventstreams for live fermentation and cold-room sensors, notebooks for the predictive models, Power BI on Direct Lake for the dashboards, and Data Activator to turn a temperature breach or a fault-risk score into an alert the cellar hand actually receives.