Personalization, Messaging and Conversational Commerce - Indix



Personalization, Messaging and Conversational Commerce

Hi! Want this? This? Too Creepy? Personalization, Messaging and Conversational Commerce is a new Y Combinator company working on AI for conversational commerce—i.e., powering commerce with Artificial Intelligence (AI)-driven messaging. I listened to a 20 minute VC podcast with the founder, Puneet Mehta, a few weeks ago while driving across the Pacific Coast Range to see my parents at their house on the Siuslaw River in the small town of Mapleton, Oregon (pop. 693). I’m obsessed with messaging as the new user interface  for everything and, at Indix, we are learning about new modalities for commerce every day.

Messaging and commerce are both highly dependent on personalization. A conversation (aka messaging) is inherently a one-on-one exploration of the other person and to fail to personalize it means that it will be no better than the customer service telephone trees of the 1990s. Commerce is equally personalization dependent. If the signal to noise ratio isn’t honed to a very fine point, there’s no compelling call to action.

My thinking on personalization and conversation began when I was listening to Puneet talk about his startup, which already has many brands as customers. But listening to my parents talk about Susie, the librarian who runs the two-room Mapleton library, really crystallized where we need to go. So I asked to visit the library while I was there.

The library is housed, along with a hair salon and a couple of other things, in the Lion’s Club building. The library used to be in a building next to Frank’s Bar, but apparently got moved because too many people were dropping their kids at the library in what I think of as a “babysitting hack” while they had a few at the bar!

When my dad walks into the library, Susie immediately hops out of her chair and with great enthusiasm chirps, “I have a book for you!” No more than 10 people a day come into the library and most are regulars (more borrow digitally—like the people who live in Deadwood—pop. 294, twenty miles upriver from Mapleton with no library of its own), so Susie has had a chance to build up quite a store of data on each patron. Susie is also married to the one USPS employee in town (Wesley) so her source of data is probably further enhanced. Susie and Wesley’s son has also helped my parents with their laptop, iPads and printer. So now we’ve got three “businesses” working together to get a holistic sense of what my dad wants to read.

Gartner talks about “digital humanism” and how important it is. The two main principles Gartner identifies are (i) that the customer’s relationship with the brand is a result of cumulative interactions with the brand in every way possible, and (ii) that the customer’s “personal space” should not be invaded (the prohibition on being “creepy” in how much you know about the customer).

What this misses is that the key for true digital humanism and true personalization is two-way relationship building. That’s what Susie has with my dad. She has lots of data about him, probably more than she “should” but he doesn’t care and would gladly give her more. She can’t really invade his personal space if she tried because the trust is so well established. As a result, he accepts her book recommendations readily.

There are lots of reasons for this, but the key is that their relationship is two-way. He knows a lot about her too, and they have had a series of conversations that have accomplished that. They’ve “messaged” extensively and those interactions have accumulated to build up mutual trust. The protected sphere of “personal space” has therefore shrunk for both of them.

This is why messaging at scale, with AI, makes sense. Granted, there’s a lot of work to do to make agents enough like Susie that it works, but already has some success (and they aren’t even at demo day yet). Puneet has his agent stop messaging if the customer starts dropping f-bombs. In response, the customers have explained sheepishly that they didn’t realize that the agent would stop messaging if swear words were used and have even apologized to the agent. Nice.

Watch for early signals on how this works, and think hard in general about how to make the discovery two-way and truly conversational, not an interrogation of the customer.

  Download the Pervasive Commerce White Paper
Get the Pervasive Commerce White Paper

Leave a Reply

Your email address will not be published. Required fields are marked *