If we onboard a client and cannot significantly improve metrics they care about we’ll lose that client. If not everyone at the client is sold on the work we’re doing then it has to be even more clear cut.
So when I see a 15% email capture rate on an opt-in popup that’s getting seen by paid traffic, with a funnel leading to a clear $80k in sales value over a year, it’s really hard for me to tell a client we shouldn’t use it next year because it maybe annoyed some percentage of the 85% of people that didn’t opt in (plus probably a lot of the people that did opt in through it).
After “earning our keep” with that $80k in value, we cannot say, “surely a utilitarian approach to minimizing annoyance across the land might magically result in higher sales.”
A/B Testing Not Being Annoying
We’ve even thought about testing that against a less annoying campaign approach. If we can match a control and be less annoying, then the less annoying variation should trump the control. But if its not an either-or situation, this can backfire and the client will end up wanting to do both.
Look at Target.
They are the best at using purchase data for predictive personalization.
There are a few holy grail turning points in someone’s life when their purchasing behavior becomes more fluid and able to be influenced. Typically they are major life events. Big B2C companies want to know those events so they can scale personalized promotions through coupons.
When Target does this at scale – it feels impersonal, almost anonymous. We get customized catalogs with coupons for products we care just enough about to bring us back through the door sooner.
Over time, it “works” but trust goes down
But over time trust goes down. With Target, in that above link, there was the famous story of how a father found out his teenage daughter was pregnant based on being sent a catalog personalized with maternity products.
When you have the economies of scale of Target you can get away with this for years.
When you’re a personal brand, not so much.
Of course, you aren’t trying to take over the world with predictive analytics, but any “leg up” that moves a needle you really care about is going to have unintended effects.
So tactics, vanity metrics, and any goals you set need to meet some quality control guidelines.
Maybe you have a checklist with things like “is this [ chatbot ] maybe going to annoy people?”
Only you can decide that, but next we’ll talk about coming up with brand centric KPIs to help balance those growth KPIs so you can stay on mission.