People Are More Than Personas: Digital Empathy Principle 13

Mike King did air quotes on a podcast. Audio only. He knew nobody could see them, so he told us.

We were about 53 minutes into part two of our catch up and I'd asked how he builds personas when every data set is skewed by somebody's ego. He started by pulling the rug out. "The bottom line is that it's never accurate," he said, and then, for the listeners: "And for people that couldn't see me, I just did air quotes."

13
People are more than personas

A persona is a starting guess. Go and meet the real people and let them prove you wrong.

"Personas are inherently not accurate, too. Personas are just a hypothesis."

Mike KingS01E03

> persona 3 was "busy mum, 34". met her. she was a dad, 52. hypothesis rejected.

Most of us have been in the workshop where a persona gets a first name, a stock photo and a favourite coffee order. By lunch the whole room is talking about "Sarah" as if she might ring back. Mike's point is that Sarah was never real. She's a story we told ourselves using whatever data we happened to have.

A hypothesis wearing a name badge

Mike compares personas to an analytics report. The report tells you 500 people landed on a page and bounced. It doesn't tell you why. "Maybe, I don't know, their computer might have turned off," he said. You're making assumptions based on what the data tells you.

A persona works the same way: "you're taking all the available data and trying to tell a story from it."

Which doesn't mean you bin them. Mike uses them to work out "what motivates this user, what is going to get them to buy", then acts on that, "and then you judge, again, from your analytics as whether or not you were right."

Plenty of teams do the first half. They build the persona, make it look nice and stick it on the wall. Hardly anyone does the second half, the bit where you go back and check. So the guess quietly hardens into a fact, and a year later the marketing plan is built on a person who doesn't exist.

There's a nice irony in this coming from Mike. Earlier in the same episode he was explaining that he wants the real story from rappers, "not the media trained story". An untested persona is the media trained version of your customer, tidy and a bit fake. (More on that in tell a compelling story.)

Start from data, not the loudest person in the room

Mike's description of how personas usually get made is painfully accurate. You get in the room, you do the affinity mapping session, "and then like the highest paid person says, 'Well, you know, this is who our audience is. I know because of X, Y, and Z.' And you're like, 'Yeah, you don't know.'"

His fix is to feed the session with CRM data, analytics, studies and surveys. "You're gonna be way closer to who this person is. And then that story you tell is gonna be way more rooted in fact."

We've written up how to build a buyer persona step by step on kwasi.com, and Mike's own persona driven keyword research process goes deep on the data side. It makes the point that what people say they do isn't necessarily what they actually do.

The people who aren't buying

Then Mike brought up what he calls aspirational personas. These are the people you're missing. "Who is this person? Why don't they buy from us? Why don't they consume our services?"

Each guess turns into something you can test. Is it that they're in a channel you're not on? Is it the messaging? Is it the layout of the page? Now you've got a list of action items instead of a laminated poster.

And because Mike ties personas to keyword research, the business case writes itself. He can put a number on the missing segment using search volume, conversion data and average order value: "Hey, I know you're getting these people, but these people were $40 million. You want to make some money or no?"

Let real people prove you wrong

Wil Reynolds makes the same case in talk to the people behind the screen: sit with a few real searchers and listen. You don't need a big research budget. Five honest conversations will poke more holes in Sarah than another month of dashboard staring.

Watch what you build it from, too. Mike's bigger worry in that stretch of the episode was bias: that our data and our tools reflect the people who made them. A persona built only from your existing customers will tell you heaps about who you already attract and nothing about who you've been putting off. More on that in you can't automate empathy.

Before your next persona workshop

Label it. Put the word "hypothesis" at the top of every persona document you have, plus one line: how we'll know if this is wrong.
Check one claim. Pick a single statement in a persona ("prefers mobile", "price sensitive") and test it against your analytics or CRM. Fix it or delete it.
Call five customers who supposedly match the persona. Ask what nearly stopped them buying. Write down their exact words.
Write one aspirational persona. Who isn't buying from you, and what are three reasons you suspect why? Turn each reason into a test.
Ask for the source. Next time someone senior says "I know our audience", ask, nicely, what that's based on.

For the full conversation, listen to Mike King part two or read the written interview on kwasi.com.