On Analytics and Empathy

All analytics does is test your empathy.

For example – if your landing page speaks the language of your customer, and your customer can tell that you obviously understand them, and can make their lives better – then your landing page will convert.

If you understood your customer, you’d know what words, or colors, or experiences would attract them.

A/B testing words, sentences, or colors, is a poor way of gaining understanding of your customer. It’s poor because it’s slow. And hypothetical. You’re guessing why your customer doesn’t like that sentence you just tested.

Another example is in keyword research for SEO. SEO marketers love saying how important keyword research is.

But, if you really understood your customer – through surveys, one-on-one chats, ethnographic research, or whatever – if you really understood their pain and how you could make it better, wouldn’t you naturally start using the right words?

So, analytics verifies whether you actually have empathy for your customer, or not. But to rely on analytics completely, is missing the point.

You’re missing the fact that your marketing must be based in empathy first, and analytics keeps your empathy accountable. Analytics and metrics are a supplement, and secondary. Understanding your customers’ pain, context, and story, is primary. Then empathy creates better tests.

Empathy also lets you understand what’s good for the customer, long term. What’s good for the customer long term, is good for your business growth long term. So, empathy stops you from sabotaging your business growth, by stopping you from making things that will not resonate with your customers long-term.

Another thought – A/B testing, or multivariate testing, involves testing one variable at a time to determine causality. If you converted better, that variable was the cause.

Problem is, empathy is non-linear. It’s both/and, not either/or. When you understand your customer more fully, you understand many variables that could influence your customer more positively. All at once, you understand which colors, words, and experiences resonate with your customer.

Sometimes it’s better to make the big leap and test all variables at once, especially when you have low traffic. With low traffic, you don’t have time to gain statistical significance on one variable at a time.

When you do it this way, it means you find out whether you’re right or wrong, but in aggregate. One or a few of many variables could be the cause, but at least you know whether you understand your customer better as a whole, or worse.

Unorthodox, but it has worked for me.

And it makes sense, doesn’t it? Wouldn’t you rather test based on your understanding of your customer’s persona as a whole, not just one fragment of their thoughts, feelings, and preferences?