Ask two collectors when a 2016 Barolo is ready and you will get two answers. Both can be right. A drinking window is not a fixed property of a wine. It is a prediction about a specific bottle, in specific storage, for a specific palate. Cellared starts from a structural base window and then moves it for the things a generic chart cannot see.
This is the personalization layer. If you want the underlying scoring model, the Cellared Ageability Index methodology documents all ten factors and where the data comes from. This guide is about what happens after the base score: how a generic window becomes yours.
The base window is the starting point, not the answer
Every wine in Cellared gets a base drinking window built from its structure: tannin, acid, varietal aging curve, body, oak, alcohol, and the rest. That base window is expressed as four points, not one year: when the wine opens up, when its peak starts, when its peak ends, and when it heads into decline.
Most tools stop there and hand the same window to everyone. Cellared treats the base window as a hypothesis and then adjusts it with three inputs that are specific to your bottle and to you.
Storage: the factor only you can supply
The single largest source of error in any drinking-window estimate is storage. A bottle held at a steady 55 degrees ages on the schedule the model expects. The same bottle kept in a warm closet ages faster, sometimes much faster, and no amount of structural scoring can know that unless you tell it.
Cellared assumes proper storage by default: around 55 degrees, stable humidity, dark, low vibration. When you record that a bottle has lived in warm or fluctuating conditions, or arrived from summer shipping without temperature control, the model tightens the window and pulls the decline point earlier. Confirmed compromised storage flags the bottle for near-term drinking regardless of what its structure would otherwise allow.
This is why two bottles of the same wine and vintage can carry different windows in your cellar. The wine is identical. The provenance is not.
Vintage: the same wine in two years does not drink alike
A warm, ripe vintage and a cool, classic vintage from the same producer follow different curves. The ripe year drinks earlier and rewards less patience. The structured year demands real time before it shows its best.
Cellared applies a vintage modifier on top of the base score, drawn from a curated vintage-ratings table covering the major regions. Where a vintage has a documented quality tier, the modifier shifts the peak window earlier or later than the regional baseline. Where vintage data is missing, or the vintage is too recent to have a stable assessment, the model falls back to the baseline and widens the window rather than pretending to a precision it does not have.
Palate: your ratings move the window toward what you actually like
The model targets a balanced peak, where primary fruit, secondary development, and early tertiary notes overlap. That is the consensus sweet spot. It may not be yours.
If your highest ratings cluster on younger, fruit-forward bottles, your real peak sits earlier than the consensus. If you keep rating mature, tertiary-driven wines highest, it sits later. As you log and rate what you drink, Cellared builds a per-varietal taste shift and nudges your windows toward the style you reward, typically by up to two years. The shift is a calibration on top of the consensus, not a replacement for it. If your palate is genuinely five years off the model on a varietal, that is a signal worth chasing yourself; the model will not carry you that far on its own.
A worked example
Take two bottles of the same Napa Cabernet, 2015. The base window from structure alone might be 2021 to 2035, peaking 2026 to 2032. Now personalize it:
- Bottle A has lived in a 55-degree unit since release. It keeps the full window.
- Bottle B came from an online auction with no storage history and shipped to you in July. Cellared tightens it, pulling the close year forward and flagging it to drink sooner.
- 2015 was a warm, generous Napa vintage, so the vintage modifier nudges both bottles slightly earlier than a classic year would.
- Your own ratings lean toward fresh, primary fruit. The palate shift pulls your peak start earlier by a year.
Same wine, same vintage, two different windows. That is the entire point.
Confidence: a wider window is sometimes the honest answer
Personalization does not mean false precision. When a wine has thin structural data, an unusual style, or a vintage too new to assess, Cellared returns a wider range and flags it as lower confidence. A four-year window means the model is confident. A ten-year window means it is being honest about what it does not know.
Natural and minimal-intervention wines are the clearest example. Zero-sulfur reds, pet-nat, and skin-contact whites age on timelines the reference data does not cover well, so Cellared widens their windows rather than guessing.
How to read and act on your window
Your personalized window still resolves to the same four points:
- Before open: the wine is too young. Structure dominates and the fruit has not knit together yet.
- Open to peak start: approachable, still climbing. Good for occasions that suit a fruit-forward expression.
- Peak start to peak end: the recommended drinking range. This is where the wine has the most to say.
- After the close year: at risk of decline. Not necessarily faulted, but the odds shift against you each year.
When a bottle crosses into its peak range, Cellared surfaces it so it does not get buried in a large cellar. For the signs that a bottle has slipped past its window, see how to tell if a wine is past its drinking window. To check a single bottle against its expected window, use the drinking window calculator.
What personalization does not do
The model is a tool, not an oracle. It does not taste your specific bottle, so cork variation and micro-oxygen differences sit outside what it can see. It does not track a producer quietly changing winemaker or style. And it will not override its own structural read on the strength of two ratings; the taste shift is bounded for a reason.
The most reliable calibration is still yours: open the first bottle of a case as a test, rate it honestly, and let both your palate and the model learn from the result. Personalization makes a large cellar tractable. Your judgment makes the final call.