Seeking advice on selecting wines for someone else

Hello Wine Berserkers!

I would love your knowledgeable opinions on a unique wine challenge I’m working on.

Together with my father, we manage a small wine ferment-on-premise (in a small Canadian town) with a selection of around 350 wines. My background is with computers and as a lead generation tool for our business we wanted to build a very simple ‘recommendation engine’ for our selection of wines. To put that simply, a potential customer could answer a series of questions and the system will provide some recommendations it believes they would enjoy. (A dumbed-down robo-sommolier if you like, just to serve as a starting point for potential new customers on our website.)

In order to create such a system, I’m focusing in on two areas:

  1. What key metrics to measure our wines by. (Sweetness, acidity, tannin, oak, ect.)
  2. What questions to ask in order to narrow down the options.

It’s on these two areas that I’m also looking for advice. Specifically, some questions below:

  • Which metrics or qualities do you personally use to help pair someone else with a wine they would enjoy?
  • In your experience, are certain metrics/qualities more or less significant in determining enjoyable options for someone else? (For example, sweetness is a good example of something that can potentially narrow the options significantly.)
  • What questions would you ask someone with previous wine knowledge to determine their preferences?
  • What questions would you ask someone with NO wine knowledge to determine their preferences? (This is a tricky one I’m struggling with! A good example of such a question is: “Do you prefer apple or grapefruit juice?” in which the answer can point to whether they would prefer a dry, acidic white or something sweeter).

I sincerely appreciate any advice the knowledgeable community has to share. I hope everyone has great holiday ‘wine-up’ planned! (That was terrible wordplay on ‘lineup,’ sorry =P)

Cheers, [cheers.gif]

Kris S

What do you consider a moderate price range. I have a friend that thinks that’s $10-$20 and a friend that thinks that $40-$100. There’s a big difference.

And for recommendations, you have to try a ton of stuff to find your palette and what you might like. And even that will change over time. It’s almost impossible to give recommendations. But to zero you in on what you like I would look for…

Shiraz from Australia. Bold, jammy stuff for the most part.

California Cabernet. Strong flavors, usually heavily oaked.

California Central Coast or Oregon Willamette Valley Pinot. More mellow flavors and less oak.

Zinfandel (pretty much anywhere) - Bold, strong, jammy flavors, but different than the Aussie Shiraz.

Bordeaux - More tannic, but still big wines.

Spanish - Tempranillo, unique flavor profile.

Malbec, Grenache - lighter reds. I would argue if you like these, you’ll probably also like Pinots and shy away from zins, shiraz, and cabs.

There are a bunch of other wines I haven’t touched on, but if you try all of these and report back, we’ll at least have an idea of what you like.

Ask what price range they’re interested in and what wines that they’ve already had that they liked. Then match similar style wines in the right price range.

Price is easy but many won’t recall specific past wines. For those, you have to be able to differentiate on style. Good luck with that. If you spend a little time here you’ll see a wide range of opinions on whether a given wine is too little, just right, or too much with respect to:
Fruit presence
Structure
Body
Sweetness
Ripeness
Acid
Tannin
Alcohol
Oak
Funk/earth
Green/vegetal
Age
I’m sure there are others.

Assuming you can sort the wines according to these characteristics, you will still face the challenge that the typical user may not understand their meaning or may have a different perspective on what is ideally fruity, ripe, structured, oaky, etc.

While I’m sure you could build a recommendation algorithm, I’m pessimistic about how well it will work unless it is based on input of known wines the customer likes. It might be better to offer tastings and make recommendations based on customer reactions.

Champagne is always the answer

Assuming you can sort the wines according to these characteristics, you will still face the challenge that the typical user may not understand their meaning or may have a different perspective on what is ideally fruity, ripe, structured, oaky, etc.

Indeed. Our terms of art would mystify most casual wine drinkers. I think you are looking for questions, not obviously related to wine, that do correlate with wine preferences. The big mega producers may have some idea of this, but no idea how to find such questions (and answers).

One other approach might be easier: to focus on what the wine will be used for and the sort of setting, time of year etc it will be consumed in.
Winter beef stew versus summer salads on the deck sort of thing.

Thank you all for the great feedback, I appreciate it. (And I got a good chuckle there cjsavino… picturing a system that asks many questions then just responds with the same phrase for everybody… “Buy Champagne”)

It is a unique challenge and I agree there is no way we can make a system that would be perfectly accurate or provide a great recommendation everytime. While I do intend to use it as a lead generation tool for our family business, each potential customer would need to speak with a human before actually making a decision, so we can curate.

That being said, I really like the idea of using price point as a way of breaking things down earlier in the process. I think that’s a good indicator not only of how experienced their palette might be, but also how complex of a wine they would prefer.

What you said David reflects my own experience. I’ve had people come into the shop and say they prefer wines that aren’t sweet… but then they’re favourite varietal is a Moscato. Interpretation is difficult, and I’m no trained sommelier, so I don’t expect a system will be able to get that right everytime.

That’s where I’m thinking you can use price point as an initial evaluation. So if someone says they like full bodied, but prefer wine <$10/bottle, then often I’ve found they actually prefer something with medium body (since they haven’t likely been exposed to well aged vintages with truly full body).

I also like the idea from Richard of choosing a wine based on the occasion it will be consumed. The majority of customers who come in the door are buying wine for a specific reason – though we always have a lot of happy faces just topping up their cellars too. So it is a mix, and each time they come in there is the potential they’ll choose something outside of their normal comfort zones. That’s where this approach makes a lot of sense.

I’ve got a lot to think about and appreciate the ideas you’ve all shared.

To keep the conversation going a bit longer, I’m curious to ask more directly: If you have a friend or guest you’re serving who has very little or no wine experience, where do you start? What approach do you take to finding something they would like? Do you base on your own preferences when not knowing theirs, or start with a mini-tasting to find where their palette lies?

Cheers!

Kris

One other idea of a different approach. Rather than try to elicit what someone might like who does not know much about wine, you could structure the conversation along the lines of, if you want something big and bold; or light and sweet; etc. I think many people might have a hard time saying what they do like. Maybe I am projecting here. I always find when someone asks what I like, that I am invariably at a loss, as I like to try lots of different things, and as Richard says, it depends on the occasion, circumstances, etc.

I do know what I generally do not like. So you could also try to use that to narrow things down. If you know you don’t like big jammy wines, then on to another set of questions.

Good luck!

One of the problems is that we tend to remember the things we don’t like much more clearly than those we do like. And the second problem is what people have said above - describing tastes is difficult and imprecise. There’s an entire thread on this site about stupid wine descriptions. The cliché is that people say they want dry but they drink sweet. That is a good illustration of the problem - they may really believe that what you call sweet is what they call “dry”. If you talk about tannin, some people will have no idea what you’re talking about.

I’ve had someone tell me that sheydidn’t like a particular wine because it was sour. Duh. It’s wine. After trying dozens, we finally clarified the fact that she didn’t like wine.

Suggesting specific grapes with descriptions becomes useless instantly. Some grapes, like Cab/Merlot, Riesling, and Pinot Noir, are recognizable from a number of locations because they always taste like themselves. Other grapes, like Syrah, Tempranillo, and Pinot Gris can taste wildly different, depending on where they’re from and who made the wine. And unless you’re obsessed with drinking only monovarietal bottlings, it’s pointless to discuss the flavor profile of any grape because it’s going to be quite different
when blended. Last night I had a Syrah/Malbec/Merlot blend. I wouldn’t pick it out as any single one of those.

That’s where I’m thinking you can use price point as an initial evaluation. So if someone says they like full bodied, but prefer wine <$10/bottle, then often I’ve found they actually prefer something with medium body (since they haven’t likely been exposed to well aged vintages with truly full body).

I don’t understand this. Not only do ageability and body have nothing to do with each other, bigger wines can seem much lighter after thirty or forty years. Price point is a bad proxy for quality, although generally if they’re drinking under $10 all bets are off. In those cases, you can probably just recommend another under-$10 wine.

I’ve seen a bunch of click bait quizzes where ‘scientists from MIT’ discern one’s perfect wine from their coffee, fruit and chocolate likes/dislikes. Needless to say, I’m skeptical.

After price, the most meaningful question that a sales person has asked me is “what food will the wine accompany?”.

Thanks Ron, Greg and Kevin for adding to the conversation.

I really relate to what you’re saying Ron, because it’s the same for me! I’m an explorer and love to try new wines (matched with the fact that I’ve only really been tasting a few years now, so there is a lot to try). So given the same questions, I would have a hard time saying what I personally do or do not like.

After having some time to think things over, I’m leaning in the direction of making a survey based on each situation. Since what someone likes in a white can be vastly different than a red, and also because they may not be choosing wine for their own tastes – using it more like a searchable directory than a recommendation engine. I’m also highly aware that no matter how it’s built, it simply can’t be perfect and it will never be more than a starting point, which will be followed by human curation.

Greg, you’re so very right – we tend to forget what we like but those negative experiences stay with us. In addition, the qualities of wine vary for each person based on their own interpretation. I do kind of like the approach of narrowing things based on what someone doesn’t want, as that will be easier for them to answer, but that approach only works if someone has enough experience with wine. I think it’s interesting your take on my price vs body comparison, as that shows me there is a difference in our own interpretations. My experience is biased, dealing mostly with wine customers who are retired and happy with low-cost or middle of the road wine (say >20/bottle on average). For the demographic I serve, I’ve found that their interpretation of body is skewed, since they’ve been buying low-quality wine for years (before us) that read ‘full bodied’ on the labels.

Based on my experience/research, body has a strong correlation to alcohol percentage, due to the viscosity of alcohol vs. water. So a 12.5% French Cab Sauv, even if it says ‘full bodied’ tends to have a lot less body than say a 15% California Cab. (Again, perhaps biased on my limited experience).

Kevin, I’ll have to look up that quiz and see if there’s anything of value there I can use! I agree though, after thinking a lot of this through, it’s a real challenge. Even if you had 10 million people using the system, and used machine learning to train an AI to pick wines – it simply couldn’t be accurate. The data would be skewed because there is no truly common standard, and so no AI could ever do the job a good sommelier can.

Thanks again for everyone’s thoughts and opinions. I love hearing your perspectives and learning from your experiences.

Cheers and a Merry Christmas to all,

Kris

Same to you Kris.

Funny but when I have people who don’t know wine very well but who like to drink it, and they’re buying cheap stuff, they almost always like whatever I serve them, since it’s better than what they’ve been drinking.

I suppose that at some level, even if you don’t know anything, you can tell a decently-made wine from something more generic. I remember the first time I spent more than $10 on a bottle - that was many years ago, when you could still buy $10 wine. But on opening and pouring the wine, we could tell immediately that it was better than what we’d been drinking. Not sure exactly what it was.

As far as disliking things - I think that’s evolutionary. If something made you sick, you specifically remembered to avoid it in the future! OTOH, if you liked something, you’d try to find more in your foraging but you’d eat whatever looked promising. No clue if that’s true or not, but it makes sense to me. So one thing you may do is ask people to describe the flavors they don’t like in wine. My Chinese customers finally figured out that they quite liked a Moscatel sherry I had, since it wasn’t so sour!

What you’re trying to do is a lot more complicated than you think. The person’s level of experience with wine has a lot to do with deciding what sorts of questions will be meaningful. Also, the way people answer certain questions (tone, etc.) can be more telling than the answers themselves. In my experience working retail, general questions about wine preferences are not usually helpful. Sure, some wine drinkers are like the people here and can tell you exactly what they like, but many/most wine drinkers are not that way. And this is all just for the people “with previous wine knowledge”, which is a very diverse group in terms of how much knowledge that is. Throw people with no wine knowledge at all in to the mix and it could take years to develop something useful. Staff members who know how to read the customer and ask the right questions can do a lot better than any software program, and I’ve seen some programs that companies have put quite a bit of money and time into and market commercially.

Hi Doug, thanks for joining the convo. Working in wine retail, I really value your opinion.

If you knew me personally, doing something near impossible is right up my alley. Not saying it will be successful, but I like to try. I also agree it will never replace a good staff member or knowledgeable human, but we’re looking at it more as a tool for our small business (just myself and my father) as a first point of contact. That’s where I know there is some risk and we may opt not to use it overall, but the challenge intrigues me.

Currently, I’m looking at building this system as a ‘one-off’ each time. No attempt to build a user profile behind the scenes as I don’t think that is done accurately until we can have a connected system with ratings, ect. So each time the ‘system’ is used, it will just provide some ideas based on what they are looking for at that particular time. It will also strongly encourage them to come to our shop and meet us humans, and with every time someone uses it we will be notified, so we can followup manually by email if desired. So I plan to have a strong human connection before anything would ever be purchased. (We don’t have online ordering, so it’s a key digital starting point for potential customers).

I hope everyone is having a great holiday so far! Cheers,

Kris

I think you on the right track with ideas - people are intimidated and want help but also want (some) choice. Hearkening back to my favorite retail experience - the owner assessed my time, interest and price range, then walked me through his (albeit small) store, stopping at certain wines to tell me about the grapes, their characteristics, and the farmers who grew them and made that wine. He didn’t ‘sell’ any of the wines during the tour but left the bottles standing so that I could easily go back.