Back to blog

Written by Ben Crudo, CEO of Diff, originally published by the Financial Post on June 5, 2018.

Be honest: Have you ever met a chatbot you actually liked? Or that actually improved your shopping experience?

I haven’t. So much ink has been spilled on such AI-powered shopping tools as chatbots and voice-enabled virtual assistants, it can seem like these bells and whistles are now the standard for any retailer.

Spoiler alert: we’re not quite there yet.

The fact is, unless you’re an Amazon-sized behemoth with petabytes of customer data that a robot brain can really help you sort through, humans still do a better job at most of the tasks that really count toward retail success. AI-forward services and experiences offered by such brands as Dominos,

Neiman Marcus and Sephora are certainly sexy, but just because a feature is invented doesn’t mean it’s beneficial for your company — or worth the investment.

Merchandising gone wrong

Let’s take something as elemental as merchandising — how products are arranged on your website. Every once in a while Amazon’s AI-powered recommendation engines will suggest a “related product” that I’m actually interested in. But they’ve got one of the biggest catalogues in the world to pull from (and even then they don’t always get it right).

For the majority of brands, it still makes more sense for actual people — the ones in touch with customers and sales — to handle critical functions like formatting your landing page. AI won’t necessarily have the same awareness of cultural trends and events as a real human will — like knowing a heat wave is going to spike sales of sunglasses, or an upcoming protest march might inspire a rush on shirts with political slogans.

Bot backlash

Ebay’s virtual shopping assistant is a fun example of the possibilities chatbots offer, but the reality is that 85 per cent of consumers prefer talking to a real person. No surprise, given the high failure rate of this tool. Facebook’s much-hyped bots, for one, can’t understand commands 70 per cent of the time.

The obvious end goal is to get to the stage where bots are measurably improving customer satisfaction — and reducing the labour involved. Right now, however, chatbots are really no more effective than a good Frequently Asked Questions page (which is to say, pretty ineffective). They can generally answer very simple queries, but anything beyond “Do I qualify for free shipping?” can lead to an exercise in aggravation.

That’s why even the best bots today need to be backed by a human customer-service team. Indeed, customer service is such an important opportunity to build loyalty — sometimes your only real touchpoint with the shopper — that it may be best to leave bots out of the picture altogether, at least for now.

Voice-assisted overload

AI-powered smart speakers such as Amazon’s Echo and Google Home have ushered in a brave new world of voice commerce — finding, selecting and purchasing products with simple voice commands. But utilizing these tools effectively for e-commerce turns out to be an uphill battle.

Not only have these technologies given rise to serious concerns about personal privacy, but retailers who go to the effort of building compatible voice apps (called Skills on Amazon’s platform) aren’t seeing great returns. Only three per cent of downloaded Skills are used again after the first week of installation.

The challenge for brands here isn’t in building the programs — it’s fairly plug-and-play — but in convincing customers to remember and use their specific commands when ordering a product. This is no easy feat considering there are literally 15,000 voice options in the Alexa catalogue.

AI that isn’t sexy, but actually moves the dial

Tech will get better (and cheaper) over time. As it does, more opportunities will emerge for AI to add real value for most retailers. Soon enough, we’ll laugh at the days when we landed on an e-commerce site and all saw the same thing— in the future, we’ll each get an impeccably personalized home page, courtesy of AI. But for now, it’s worth only using what actually works. Don’t fall — or pay — for the hype.

Back to top

We'd love to chat