The data that remains: managing the information wine produces is the real strategic value

A bottle of wine is not born in the cellar. Nor is it born during fermentation. It is born much earlier, the moment someone decides which vineyard to plant, where, with which variety, which rootstock, which training system, which planting density. It is the first strategic decision of the entire chain, and it is also the most important — because it constrains the decades of production that follow.

Every decision that comes after — agronomic, oenological, commercial — depends entirely on the moment a vineyard is planted. This first choice is not made on instinct: it too depends on data and specific information.

Today we are not going to talk about the considerations involved in establishing a new vineyard. We want to talk about a topic that is central to LoraWine: DATA.

The real perimeter of wine data

When people talk about "wine data," the conditioned reflex is to think of sensors, drones, consumption metrics and analytical parameters. It is a partial and misleading representation. Wine data is much broader and far more heterogeneous:

• Instrumental data from microclimate sensors, weather stations, vigour maps, fermentation thermometers, oenological probes and automation systems;
• Laboratory data, from soil analyses, to seasonal leaf analyses, must analyses, all the way to the polyphenolic and aromatic profile of the finished wine;
• Notes and observations from the people who live the vineyard and the winery — the agronomist's note, the winemaker's comment on how a particular tank is behaving, the organoleptic evaluation;
• Process data: maceration times, fermentation temperatures, pump-overs, rackings, choice of wood, length of ageing;
• Business and management data: labour hours dedicated to a specific product, materials purchased, targets achieved, operating margins;
• Commercial and market data that close the cycle, telling us how that wine has performed.

This heterogeneity is not a detail. It is the heart of the problem.

The academic literature has framed the issue clearly. Year after year there is a growing wealth of data coming from each of these sources, but the problem is that these sources scatter easily and are often poorly accessible. The lab data sits in a PDF on a consultant's computer. The agronomist's note sits in a notebook. The weather station data sits on a vendor's platform. The cellar log sits in a management software that only talks to itself. The market data lives in a spreadsheet and in the head of whoever wrote it. When these systems do not communicate, the actual value of all this information is lost, and what remains are scattered fragments of a story that no one ever reconstructs.

Data has a longer lifecycle than the decision that generated it

This is the point the industry tends to underestimate. Every piece of wine data inevitably carries a sense of immediacy: the sensor reading is used to decide whether to spray today, the grape monitoring serves to time the harvest, the winemaker's observation guides tomorrow's racking. All of this is correct.

But if that data is properly managed, it takes on a completely different strategic value, one capable of generating a lasting competitive advantage. The technological innovation offered by digital tools is not only the ability to perform advanced calculations and analyses. Digital environments offer the ability to archive, structure and catalogue the vast amount of information that lies behind every bottle, so that it can be easily consulted and so that valuable insights can be extracted from it. The principle is codified in the literature under the name FAIR data — Findable, Accessible, Interoperable, Reusable. In the era of big data and omics, good organisation, management and description of experimental data is crucial for producing high-quality datasets, exporting robust results, making data more easily available and unlocking the enormous potential of reuse. This is not an academic exercise. It is the condition that makes the data you collect today — instrumental, laboratory, observational — still usable ten years from now, by the person who will have to make a replanting decision.

Technology should not be used to predict the future or automate decisions. The real added value technology offers is the ability to access information and interpret it appropriately for every decision that needs to be made, taking into account the greatest possible number of factors and scenarios.

The vineyard-winery link is where strategic value hides

A single agronomic choice and a single oenological choice, taken in isolation, are operational decisions. They become strategic only when the data that generated them is connected — in space (the same parcel, the same wine) and in time (multiple consecutive vintages).

The most thoroughly documented international case is the database built from the Amerine-Winkler report (1944), digitised and published in Scientific Data in 2022: harvest dates, must and wine quality, and tasting notes for 148 cultivars across five climate regions of California, integrated with meteorological data from 1911 to 2018. Note the composition: instrumental data, laboratory data, and tasting notes. Exactly the heterogeneous mix we are talking about. Eighty years later, that data is guiding replanting decisions around the world.

In Italy there is a paradigmatic example, geographically closer: a study published in OENO One analysed the historical phenological and environmental series of two vineyards in Piana Rotaliana (Trentino) from 1986 to 2022, showing how the understanding of the historical trend of functional processes — phenology and ripening — together with specific weather events, is a fundamental building block for defining adaptation strategies through genetic improvement or vineyard management. Thirty-six years of observations. Each, taken individually, served to decide when to harvest. Aggregated, they redraw the map of what is possible in that territory.

Below are some examples of questions that require this connection and that cannot be resolved otherwise:

• Which combination of variety × rootstock × site × soil × management strategy, on historical climate data, has shown the lowest qualitative variability in climatically adverse vintages? An answer that guides replanting choices and the positioning of premium lines.
• In the vineyards where I produce wine X, are the conditions in place to carry out the operations I have planned in order to increase revenue and reach my oenological targets?
• Given the conditions normally found in my vineyard, the vegetative behaviour and the territorial characteristics, which operations have proved most effective in reducing the alcohol level in the wine?
• This year's production has been low and the analytical parameters of the must/wine are anomalous, even though the general weather conditions have been favourable. Why?
• Have the tanks that fermented grapes coming from parcels with a particular leaf nutritional profile shown more regular fermentations? An answer that links vineyard fertilisation to yeast management in the cellar.
• Given the temperature trend I recorded during the second phases of ripening, can I keep applying this agronomic strategy to obtain the desired polyphenolic profile, or do I need to change my approach?

None of these questions can be answered with today's data alone, and none can be answered with data from a single domain. They can only be answered with data from the vineyard, the cellar, the laboratory and from observation, collected over the years for operational reasons and preserved in such a way that today they can still be queried.

This is where AI comes in — but not the way marketing tells the story

The technical ability to query heterogeneous sources is one of the areas in which recent AI models have made possible something that was prohibitive five years ago. A PDF of an analysis, a sheet of vineyard observations, a cellar management software, a text file, an Excel file and a historical data series are no longer perfect strangers speaking different languages. For the first time, we are in a situation where a properly designed AI is able to extract clear information from a messy pool of company data, returning patterns and correlations.

However, AI only works on data that has been correctly stored and properly catalogued. If the laboratory PDFs were saved with inconsistent names, if the winemaker's observations were on a notebook that has been lost, if the sensors changed vendor every three years with incompatible formats — no model can recover and process that data. The value of AI as a consultation tool grows in direct proportion to the care with which data has been collected, labelled and preserved upstream.

In other words: AI makes the discipline of data collection more urgent, not less. Because the tool that can finally extract value from what you have set aside now exists — on the sole condition that you have set it aside in an orderly and properly classified way.

What this means, concretely, for a wine business

Three operational principles that hold regardless of the size of the business.

First: Standardise the unit of observation. Each individual vineyard parcel must have a unique identifier that ties it to all the data that concerns it — soil analyses, treatments, sensors, observations, harvest, and then must, tank, analysis and ageing. Without this link, the cellar data and the vineyard data remain two strangers unable to communicate.

Second: Treat observation as data. The agronomist's note, a photo of an event in the field, the winemaker's comment, the tasting evaluation are not mere "notes". They are data in every respect, and they must be preserved with the same care as instrumental data — linked to the same parcel or tank, catalogued and archived.

Third: Choose tools with the decade in mind, not the vintage. A change in nomenclature three years from now destroys data comparability. A sensor vendor that goes out of business and takes the data with it leaves a gap in the historical series. A management software that does not export in open formats is a trap.

The paradigm shift underway

The international wine industry is recognising this paradigm shift. The OIV (International Organisation of Vine and Wine) has placed the digital transformation of the sector at the heart of its five-year strategic plan, bringing together experts in artificial intelligence, smart vineyards and blockchain to guide the transition. This is not an exercise in operational efficiency. It is a paradigm shift in what it means to run a wine business.

The difference between a company that uses data to react and one that uses data to decide is not measured in the quality of a single vintage. It is measured in the resilience of the company over time: in the choice of the next vineyard to plant, in the agronomic management to adopt, in the next investment in the cellar, and in defining market and marketing strategies.

To be sustainable, that choice must be based on the data collected over the years.

Where to start structuring your data? We've summarized the practical steps in a 7-point actionable guide, available for free download below.

---

Sources

1. Four decades in the vineyard: the impact of climate change on grapevine phenology and wine quality in northern Italy, OENO One, 2024 — https://oeno-one.eu/article/view/8083
2. Bai, H., Gambetta, G.A., et al., Historical long-term cultivar×climate suitability data to inform viticultural adaptation to climate change, Scientific Data (Nature), 2022 — https://www.nature.com/articles/s41597-022-01367-6
3. Grapevine and Wine Metabolomics-Based Guidelines for FAIR Data and Metadata Management, PMC/NIH — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8618349/
4. Towards an open grapevine information system, PMC/NIH — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5120350/
5. OIV, Digitalisation, the future of the vine and wine sector — https://www.oiv.int/en/oiv-life/digitalisation-the-future-of-the-vine-and-wine-sector