Don’t take it personally!

Passing through middle age, I’m fairly confident that by now I know what I like. But who else does, and how do they know it?

I would wager lots of people know my favourite film, band or game. Many will know how I like to take my rum. But how many people know the answer to all of these questions, is there anything connecting them, and more importantly, what does it say about me?

Does the data we leave behind actually paint a picture of who we are, and what we like?

Recently I read about a menswear start-up looking to recommend clothes based on a user’s Spotify data. Now I am not the biggest clothes shopper in the world, but I have been a daily users of Spotify for over 9 years, so they certainly have a good steer on my music tastes. Intrigued, I signed up.

The hidden magic

The onboarding was a simple process; sign-up with Spotify, answer a few lifestyle questions, enter my measurements. Whilst the signposting was minimal, with no explanations given as to why the requested data was pertinent, none of the questions seemed too superfluous, challenging or intrusive. A relatively quick and painless process and I was in.

And there were the results, 25 looks just for me, supposedly picked to appeal based on my music tastes. Let’s just take a look at some of those top artists.

(Kim Deal, Otis Redding, Bob Log III)

What do these artists say about me?

So I like, no, love The Breeders. Their sound, their approach to recording, their very presence sings to me. Kim Deal rocks my world no matter the mood. But does anything in their output influence the way I dress, well not that I can see. Though I will come back to Kim’s twin sister and fellow band member Kelley later.

Moving on to Otis Redding and we might be on to something here. It’s true, I always wear crew necks, never a v-neck. Maybe there is something in it after all, though I doubt they have added image classification to their algorithm.

But wait, who’s this. Bob Log III. I can honestly say I don’t think I will ever wear a glittery jump suit whilst at the same time (or at any time) wearing a motorbike helmet with a bakelite telephone receiver glued to the front of it.

Had my top Spotify artists been The Specials, The Jam and The Selector (bands I do love) then the algorithm might have been able to correctly guess I have 5 black and amber Fred Perry’s in my wardrobe. But like most people, I don’t just listen to one narrow genre of music.

So, what was I recommended?

25 trending looks, tailored to me, for my perusal and approval. Now it’s a bit unfair to judge, as everything the particular retailer is selling is definitely more towards the smart side of smart casual (with the price tags to match). But to my eye, I was given a series of generic outfits which whilst some had similarities, you certainly wouldn’t say they followed a theme, or had an obvious influence.

Oh, and the only sweater on show? A v-neck.

Whilst I honestly can’t see how my Spotify data could influence the recommendations in any meaningful way, I thought it worth trying again. So I re-joined, answering the questions in the same (honest) way, but this time using a colleagues Spotify account.

Whilst both Caroline and I do share some favourite bands (Led Zeppelin, Nirvana), there is a huge amount of divergence. She loves her pop-punk classics (I like my punk without the pop) and she can’t abide Kim (which pains me deeply).

Can you guess what happened? 25 fresh looks, but with lots of overlap and again, no rhyme or reason (or explanation) as to why things had been picked. Out of the two collections presented, I liked a similar amount of recommendations in each.

Can clothes and music tastes combine?

Actually, in a way they can, and even led to a recommendation. I said I would return to Kelley Deal (guitarist in The Breeders, R.Ring and all round rock goddess) and here we are. Whilst not her day job, Kelley loves knitting and crafts, and makes some very nice scarfs.

Hand made in very limited quantities they are hard to get hold of, but one day last year, I managed to get one of a drop of six. A couple of weeks later and a package from the US arrived.

A postcard, to me, from Kelley Deal (swoon)

I was delighted with the scarf, of course, but I had also purchased a couple of her singles. One of which was a double a-side with another band Kelley loves (she’s since recorded another EP with them), Protomartyr

One play and I was hooked. I’ve since seen them twice, and will again this weekend. Now without Kelley’s recommendation (and endorsement) I may have heard them sooner or later, but it was this personal recommendation and collaboration which led me to discover them and fall in love.

Spotify itself makes recommendations based on users listening habits. Their Discovery Weekly automated playlist has transformed my music discovery. It’s not flawless, but the amount of new music and bands I have found over the last few years is amazing.

Whilst the exact science is still a bit of a mystery, we do know that Spotify use a combination of techniques to determine what it thinks you will like, based on your listening history, playlists, and importantly, those of its other 170 million users.

They have data which is not only adequately large in size, but actually relevant.

If we were both to love 10 of the same bands, its likely you could recommend me another. But what do you think the chances are we would dress the same, or like the same food, or want to drive the same car?

I’d wager not very high.

Personalising services can lead to greater engagement and customer satisfaction, but we have to base those recommendations on meaningful data or insight. And to be that it really should be relevant.

Hooking into Spotify may seem like a novel idea (and generate column inches), but if it tells you nothing relevant about your users, then what’s the point?

The right way?

Asking users questions about their interests is still a reliable data gathering method for personalisation, so too is their data footprint, but both only help when they are relevant.

A users location can be useful when serving the weather. Knowing where you work or live helps Uber suggest destinations when you open the app (throw in the time of day and it gets even better).

Ultimately, whether the data has been scrapped from a footprint, detected from a device, or specifically given by a user, it can be used to tailor content that’s more relevant. But only if it’s the right data.

If you want to know how TheTin can help you engage your user’s then get in touch. But if you just want a recommendation, then Dick Stusso’s In Heaven is my album of the year.

As your brand and technology partner, we’ll help you discover what’s possible.

We’ll make sure that the way we work is the right fit for your business, and we’ll ask the right questions to make sure you’re set up for success.

We can help build your brand through technology, email info@thetin.net