Stuck in a wine rut? These are the best new grapes to try

Telegraph

19-03-2025

Do you ever stretch the comfort zone of your regular wine choices with a stroll into the wilds of say, white wine from Uruguay, or lambrusco – the sparkling wine of Emilia-Romagna?
It used to be up to sommeliers and supermarkets to coax us out of our wine ruts. Sommeliers combine a deep knowledge of their list with individualised recommendations that might go something like this: 'You mentioned that you like dry whites, have you ever tried Ribolla Gialla? It's a grape found in the mountains of north-east Italy and over the border in Slovenia and it's very fresh, often with a faint tinge of pithy bitterness and tannin.'
Then you could try Borgo Conventi Ribolla Gialla 2023 Friuli Venezia Giulia, Italy (12%, Booths, £14.25), which tastes of lemon curd, pears and hay.
Supermarkets, meanwhile, use ranging to confer trust. Back in the dark ages of 2009, or thereabouts, Tesco put a grape called Piquepoul into its finest livery in the hope of encouraging shoppers to switch from pinot grigio, a move so successful it transformed the fortunes of the small, southern French region in which it's grown and turned the grape into a staple. More recently, ranges like M&S Found, Sainsbury's Discovery Collection and Waitrose Loved & Found have turned obscure grapes into heroes.
In 2025, of course, we live in the era of the algorithm and 'an increasing desire for personalisation,' as Wine Society CEO Steve Finlan puts it. We want to know not just that Bulgarian pinot noir is a winner, but that it's right for us. And we act on those algorithmic nudges. When personalised recommendations, derived from AI data-mining, were added to The Wine Society's website, there was a 'double digit increase in click-throughs.'
Such methods create a more nuanced recommendation landscape than averaged star ratings. 'Some wines can be quite polarising,' says Matt Smith, buying director of Naked Wines, which has been using personalised recommendations based mainly on customer ratings, for some time. 'For instance, dry sherry scores very low with some customers but very high with others and that group also has a high repurchase rate. I like to keep wines like that in the range, it's part of the rich tapestry.'
An interesting feature of the Naked algorithm is that it takes into account categories wine buyers have assigned to each wine, such as Smooth Red or Crisp White.
Taking the taste of the wine into account in this way brings the system closer to acting like a sommelier. The US-based company Preferabli says it can push this all the way, with software that uses AI to act 'like a human expert'. Last summer I attended an AI conference at which its CEO Pam Dillon explained that tasting experts codify up to 800 different characteristics of a drink to build its profile in the software. She told the somewhat divided room: 'Our goal was nothing less than to taste through the entire world of wine and spirits one bottle at a time.' A system designed by Preferabli has just gone into use at M&S, online and in 20 food halls.