Excessive technology to low-priority situations
(Many friends sent me this excellent essay from Dan Grover on bots vs. apps— here’s my two very cheap foreign cents)
In Windhoek, Namibia, one can buy data plans from what looks like an arcade game, in what amounts to a 7-11. At least, I did. Unclear as to which option to go for (and a little nervous about missing the overnight bus to the Zimbabwe border), and feeling a bit gamble-y, I went for the dead-in-the-middle option. SIM card popped in, a pop-up overriding everything else appeared, then we were in SMS mode for what plan I was signing up for. I was genuinely impressed at the no nonsense smartphone version of the lo-fi top-up.
While choosing a Data Plan (IVR-style and in English, thank goodness), it was also the first time I came across the pop-up “Facebook and WhatsApp are Free” — the Free Basics. A source of much debate on Net Neutrality, and cracked down upon (most recently in India), it was the first time I actually had encountered it. I didn’t blink, and took the Free Basics.
I eventually ended up in Lesbos, among other refugee camps in Africa, Middle East and Europe. To be clear: nobody is downloading a custom app as they’re fleeing civilian slaughter, not to mention the zero mobile coverage (sorry everyone who did a hackathon around new apps and websites, there’s just no time and space for that when it’s bombs away). There’s a reason why many have been called Facebook Refugees, hanging on to the Turkish cell network to ask for help before the coverage is lost once on arrival in Greece.
We often forget how much Data Matter$ for the unconnected. When your app (let’s not kid ourselves — mega platform) is basically the reason why my smartphone is still a phone, that’s simply the only thing I will use.
It’s not just how heavy the app is on my data credit, but also about how much data you can collect simply and store it. To be honest, I’ll probably take the multiple taps over a pizza app any day. I don’t know about you, but when I make a pizza choice, I don’t change it that often when I want pizza delivered again. Same with takeout Chinese, really, so much so that most conversations go like this:
“Ah hello you this #7?”
I’d just like this bot to remember this pizza choice, and I can just send pizza1 or p1 — maybe I’ll look at an occasional offer from push notifications on said p1. If this can already be done while playing Call of Duty, I don’t see why it can’t be done over a specific SMS bot on the smartphone. I don’t really need the chat bot to make niceties about my day and the rain. This is not the time for delight.
When it comes to bots, backend simplicity matters more than front-end simplicity of use.
What frustrates me most are the NLP type questions. One of the first startups I was involved with many years ago was an MIT NLP powerhouse — and we were teasing out intent and decisions from crawling healthcare forums. The amount of data cleanup we had needed to do then was, to put mildly, intense. Folks talk about diabetes medication and how it affects them walking their dogs, while coping with a sick parent, for example. To this end, bots and intent are very far apart — which is why context is critical in specifying exact intent.
When I top up my phone in a not very connected part of the world, I expect few options, and frankly I appreciated the inability to do anything else until the matter is solved in the absolute most lo-fi of manners (YES 2, 2 FOR INTERNET). The only company I am currently aware of who has been acquired in this area is Kit CRM (michaelperry!) to Shopify: Kit focuses solely in marketing for your online store via the SMS-equivalent of IVR choices. It’s awfully slick.
The last frustrating point is the way the apps and bots are often discussed in a product context in San Francisco/Silicon Valley since F8, with not much thought to the wild: as though everyone is David Attenborough who happens to know nothing about nature. No one needs to go full on what refugees need from their smartphones in the Aegean, but checking out what a working class black mom in Atlanta does need may really well help that bot/app strategy. I think that’s what Kit got right, and I suspect the most successful (and useful, and a pleasure to use!) bots will be the ones who get this focus right.