Since Facebook launched their bot platform at F8 in April 2016, the press has been raving about the potential of chatbots, a new paradigm, a potential successor to mobile native applications.

There are indeed arguments in favour of a chatbot play, especially on Facebook Messenger. The platform has now 1.2B monthly active users (from 200M three years ago), which puts it on par with another Facebook property, Whatsapp, which may also consider adding bots to its features list. Compared to the typical app download flow (through the Appstore or Google Play), chat bots offer a much reduced friction, with conversations just one tap away, without the need to download anything (you could say the same about web apps). No further authentication needed (we’ll get back to this), since users are logged in (locked in) the messaging app.

So it’s easy to understand why this new exciting opportunity has prompted a gold rush frenzy among developers and investors. What if chatbots could become the new native apps? What if you could create a new ecosystem as successful as the Appstore, launched in 2008 (with some caveats as far as long tail app makers are concerned, in a winner-takes-it-all market)? Phil Libin, the founder of Evernote, who joined a fund very active on the scene, even says that the first bot IPOs are 2 or 3 years away.

The 1B dollar question is: can you build a sustainable business piggybacked on someone else’s platform? You could argue that Zynga has been pretty successful in doing so, owing their meteoric rise to some viral tricks in the early days of the Facebook feed. But its shares fell off a cliff soon after the IPO, when Facebook and Zynga amended the terms of their partnership. The company still boasts a comfy market cap of $2.55B as we’re writing these lines but it will probably never recover its pre-2012 value.

chatbot strategy

There’s an element which still played in favour of Zynga and other social gaming companies compared with the meltdown bot startups could face if they happen to have a major dispute with Facebook (or their SaaS partner): user data. All social gaming apps required a Facebook login which entitled app developers to retrieve essential data from Facebook, including a key persistent identifier: the email address.

Facebook bot developers don’t get default access to Facebook users’ data. It’s not automatically passed when a user starts a bot. The frictionless beauty of the onboarding process also happens to be the worst long term threat to bot developers if they don’t find ways to enrich their data. To be honest, it’s a bit awkward to ask a messaging app user for his email credentials. Why would you need my email (or my phone number) if we’re already freely discussing on this messaging app? We can just keep on chatting.

Just imagine that you would spend $1M on chatters acquisition on Facebook (via Facebook feed ads). Say you’re a good performance marketer, getting 1M users with that investment. You’re having an active discussion with these 1M users, via Messenger. You’ve even managed to find a way to monetise some of those conversations (a tiny fraction of the users are buying products recommended in chats). Your investors are pumped up, you’ve got a thriving community. Yes… but… one day, Facebook changes its Terms Of Service and some critical parts of your user experience are affected. You lose half of your user base overnight. The issue is that you can’t try to win anyone back, since you don’t own any user data. You’re just granted temporary access to someone else’s users (which you paid for, remember the $1M advertising spend).

And that’s just one aspect of the risks you’re facing. If you’ve used a SaaS to develop your bot’s “user base”, e.g. ChatFuel or ManyChat, take a deep breath. If one day you decide to disconnect your Chatfuel bot from your Facebook page, because you think that some of the features offered by ManyChat make more sense (and you’d be right to do so for some use cases), the userbase you’ve built via ChatFuel, i.e. the subscribers to whom you could send broadcasts, will be lost (L-O-S-T), simply because you can’t export a user base from one SaaS provider to another one. Bummer! So if you’ve invested $1M in customer acquisition for a bot operated via a SaaS provider and the startup goes bust or sells out, you’ll have to be very creative.

We’re not saying that you shouldn’t experiment with bots and find a way to integrate them in your marketing strategy, but always remember that you should develop a user base which will resist any platform policy changes and stand the test of time. This is why we recommend platform-independent identifiers, such as the email and/or the phone number, enriched with user data which you will store at your end, not on someone else’s stack. This is the best life insurance policy for your venture.

Suggestions