Can you personalise wine selections? Should you?

This slightly rambling session of the wine marketing masterclass looks at the ultimate form of targeting - the individual. And then tries to answer the question - ih and how we should target our wine

It’s Wednesday so we have The Wine Marketing Masterclass. All previous sessions are in the (paywalled) archive and most new sessions are for paid subscribers.

Remember, this is a COMPLETE MBA course in wine marketing. For just $10 a month. (A full wine MBA costs up to €30,000) you’ll discover sell more wine, more profitably, to more people. And to catch you up we’ve already covered a LOT, with multiple sessions across Market Orientation, Market Research, Segmentation now we’re in Targeting. From next week we’ll cover Positioning before getting into the 4P’s.

(Are you a wine company and you have specific marketing challenges? Remember, I’m available for bespoke consulting across marketing, exports, sales strategies and more too.)

We’ve already looked at why and how you can target big and small groups of wine lovers. And if you should target them at all. But what about the other end.

Should we tar­get in­di­vidu­al cus­tom­ers? 

Digi­tal tools and e-com­merce give you the ability to tar­get in­di­vidu­als. Kind of. And you may be thinking surely that's the best ap­proach. The dream scenario. Per­son­alised, one-to-one mar­ket­ing of wine PERFECTLY matched to someone’s palate.

Well, Prefer­abli is a Ma­chine Learn­ing-based tool that does just this. It helps wine, beer, and spir­it brands tar­get their products to in­di­vidu­als. The targeting is based on the qualit­ies of the product, as assessed by an extraordinary panel of MWs, MS’s and other experts. And targeting is based on the at­trib­utes about the cus­tom­er, as assessed by digital technology.

In a White Pa­per – you can get a copy here – the co-founder of Prefer­abli, Pam Dillon, dis­cusses how mar­ket­ing strategies have been used with­in the wine and spir­its busi­ness. She talks about the tendency to pro­mote a lim­ited set of the same products to all cus­tom­ers. And indeed companies do. She also talks of how mass mar­ket seg­ment­a­tion tar­gets large groups of people, based on just a few char­ac­ter­ist­ics.

But then along comes re­com­mend­a­tion soft­ware. This means (or at least promises) you can pro­mote dif­fer­ent products to dif­fer­ent people. You are the beneficiary/slave of these products even if you don’t know it. Every time you use Amazon, or Spo­ti­fy, or... well, any plat­form that recom­mends products to you. These serve up choices by a re­com­mend­a­tion engine. Books, Instagram posts, cosmetics, shoes, ads for household items. They can (I say “can”) be very clever too. I didn’t know I needed that last pair of trousers. But I bought them.

Wine and spirits are chal­len­ging for recommendation engines though, says Pam Dillon, be­cause there are a lot of products. And those products tend to be com­plex. In her ana­lys­is of the in­dustry Pam Dillon sees a prob­lem:

Mass mar­ket seg­ment­a­tion in the wine and spir­its in­dustry has taken the shape of  per­so­nas us­ing char­ac­ter­ist­ics such as age, know­ledge, in­come and gender. This  tech­nique is defin­i­tion­ally not fo­cused on one-to-one per­son­al­ized ex­per­i­ences. Its  prin­cip­al draw­back, apart from the in­trins­ic in­ab­il­ity to ad­dress an in­di­vidu­al’s taste pref­er­ences, is that it ig­nores life­style and con­text.

I totally agree. And we’ve seen that theme a lot in these masterclasses. For Dillon and the Prefer­abli team, the an­swer lies in Ma­chine Learn­ing tools. These will work with the feed­back that each user provides. That could be ex­pli­citly through rat­ings or oth­er sentiment in­dic­at­ors. Or im­pli­citly through the search­es you make, bid­s you place, click­s or what you actually end up buying. It can even be the “linger time” you have on a particular page or picture.

The system looks a bit like this… 

Recommendation sys­tems ana­lyse both user and product data. The image above is a general idea, and not specific to wine or any particular company. They (should) mean you get rel­ev­ant selections and recommendations. This (should) mean you deep­en cus­tom­er loy­alty and increase sales. That (should) also mean a re­tail­er can serve up options that are not just from their biggest sellers, but high­er mar­gin op­tions too that customers will buy because they’re a better recommendation. Or from smal­ler pro­du­cers. This means we can all mine the “long tail” of choice, be­cause the sys­tem knows what our customers will like.

So, does it work?

I’ve used the word “should” a bit here. This is not a critique of Preferabli or any other system. I’ve worked on developing this sort of system. And there are thousands of users who love the recommendations from Preferabli and other systems. That is testament on one level that they “work”. But wine sales are declining, which suggests they’re “an” answer, not “the answer”. I’ll have a (free to all) post looking at why on Monday. It involves a detour into Symbolic Interactionism because… of course it does.

So come on Joe, you’ve faffed about, tell us if, why, how we should target our customers

We have THREE po­ten­tial solu­tions to the question of targeting: