Okay folks, maybe cool it with the messenger bots.

There’s a small co-working space in Helena, Montana that has a more effective Facebook Messenger strategy than most marketers. They use an incredibly sophisticated neural network combined with cognitive processing to analyze natural language messages coming in from Facebook. They process the query, comparing it to past experiences, and then write a friendly, personalized natural language response.

It’s a truly remarkable customer engagement experience:

The technology powering this is (drumroll, please) a man named Matt Gorecki. He had the novel and remarkable idea of responding himself, as a real human being. I know what you’re going to say, “One-on-one responses just aren’t scalable!!”

Matt does have it easier than many marketers because he only gets a handful of messages per week. Some of us get many, many more than that already and we want that number to grow even further.

Listen marketers, I get it, the Holy Grail for digital marketing is to scale human interaction. But your customers do not care even one tiny bit about helping you find your Holy Grail.

That doesn’t stop us from trying, of course. Marketers have been so persistent in our pursuit of “scalable marketing” that our aggressive tactics have actually given rise to an entire new market of products & services aimed at stopping us — from TiVo to ad blockers. And the business of stopping marketers from doing their jobs is booming. Ouch!

Kipp Bodnar, the CMO at HubSpot, recently shared that the HubSpot marketing team is seeing, are you ready for this? 4x conversion rates on Facebook Messenger vs. email. The opportunity here is massive.

So when marketers first heard that Facebook was opening its Messenger app to automated applications, we predictably began to drool, having visions of our set-it-and-forget it messenger bots bringing in customers while we watch. Let me show you why these visions are unattainable.

To do this, I’m going to pick on Politico, but not because it’s the worst offender. They’re doing what every marketing team should do — experimenting. Politico is just the most recent example I’ve encountered (Sorry, Politico!)

Politico’s bot is a glorified “Choose Your Own Adventure” story, where they send you a message about whatever they damn well please and then give you a set number of canned responses to progress through the story. Functionally, it’s little different than an email subscription with some “would you like to know more?” buttons embedded:

Marketers suffer from a chronic condition of all-about-me-itis. We expect prospects to fit themselves into our process, rather than fitting our process to them. Messenger bots are, currently, no exception. Rather than scaling a human experience, marketers build a series of if/then statements that we beat customers in the face with until they give in or give up.

And this is exactly why Matt from Montana’s Facebook Messenger strategy is better than most of ours.

The world doesn’t need another mediocre marketing experience. Consumers don’t need another channel through which marketers talk at them. Until we have the ability to deliver a valuable, conversational experience that’s not rage-inducingly primitive, follow these 3 rules to guide your Facebook Messenger strategy:

1. Solve for the customer

I have repeatedly threatened/offered to tattoo “We Are Not The User” on the insides of the eyelids for engineers who have worked with me (don’t worry, our People Operations team told me I couldn’t). But really, marketers would benefit from the same tattoo.

Marketers need to be as obsessed with the prospect’s job-to-be-done as good engineers are with their users’.

When it comes to messaging apps, this means answering questions like:

• Why do your customer’s use it?
• What do they want from it?
• What makes messenger apps a different experience than email?

Don’t build The Thing until you know if and how people are going to use The Thing. In other words, learn a little from the Lean Startup methodology.

One of my favorite tactics to do this is the “Wizard of Oz” approach. Your starting point is not a sophisticated-but-still-not-good-enough machine learning algorithm to power your bot, your starting point is understanding which use cases will drive the most value for your customers.

The Wizard of Oz approach appears bot-like, but is really just a human responding — Matt from Montana style. In fact, having a corpus of interactions between real, human customers and real, human employees may actually improve your eventual efforts to scale using machine learning.

So my advice for most marketers considering creating an automated bot because it’s the latest cool thing and the CEO wants to be able to brag about how innovative they are — don’t. Start by responding promptly to incoming messages from any channel. Then start encouraging more people to message you. See what, why, how, when, and to solve what problems the messages are coming from.

2. Stay human, if you can

Even once you’ve figured out the perfect use-case, resist automating too soon. The programming powering most bots is still pretty dumb. Most bots are actually less sophisticated than modern email marketing automation, because they’re usually not tied to a CRM so they’re unable to pull behavioral data from things like website visits or past orders to create a relevant and personalized experience.

Right now, the experience delivered by humans is via messenger apps is just exponentially better, and at 4x conversion rates, maybe the little extra TLC is worth it?

3. Stay tuned to where chat is heading

Did you ever notice that in science fiction movies no one is typing out messages to interact with the computer? They’re talking. It’s because speaking comes far more naturally to us than writing and is obviously a superior form of interaction.

Let’s be honest, this rant would be far more valuable for both you and me if it were being done by voice in a one-to-one interaction.

Voice rules and text drools.

Keep that in mind as you experiment with chat. It’s possible that in the very near future, many or most of your chat interactions will move to voice. This additional convenience might upset your chat strategy, but your customers will love it.

Bottom Line

Sometimes doing it poorly is worse than not doing it at all.

If you don’t have the time to be Matt from Montana and your “chat strategy” is just to implement some generic bot technology and hope for the best — maybe just don’t.

For some reason, while writing this rant, I’ve had this line from the Star Trek: Deep Space Nine episode “You Are Cordially Invited” stuck in my head:

We might not yet be able to deliver the dreamy, automated convenience that we want — but in the meantime, let’s not use such obvious fakes.

“Computer, please ask the reader to click the ❤ to recommend this article if they enjoyed it so more people see it. If they didn’t enjoy it, ask them to click the ❤ so more people can see how horrible it is and avoid making the same mistake.”

… yeah, typing that is way less fun than speaking it.