Before you “dipset,” (that is, bail) scan the bold items below.
Optimize like it’s a landing page, not a mass mailer: We can talk all day about A/B testing, multivariate and machine learning, or what subject lines work, but unless you change your thinking, you will never produce double digit improvement in what you do.
When you optimize landing pages, you tend to think in terms of experiments and optimizing as traffic comes. When you batch mail and are racing to get to the inbox at 8 a.m., you limit what you can do.
Real test optimization comes from testing things perpetually in motion, building experiments that take variants (e.g. subject lines, body copy, offer strategy, creative treatment and even ISPs behavioral patterns) and allowing small samples to dictate optimization.
I just don’t believe that marketers know enough about what to test, have a historical view of what worked, and are creative enough when in crunch time. You have to let machines do some of this for you. The variants aren’t hard to determine, but don’t expect a batch campaign on Monday to tell you what will work on Thursday or next Monday. It doesn’t work that way. You have to test iteratively in smaller samples with ongoing consistency across triggers, automations, transactional.
Take 80% of what you think you know, and retest it: We live in a world where a good outcome is that maybe 40% of our audiences sees the ad/email, and maybe 20% of those will click on it and maybe 25% of those will do something past that first page they land on.
Many things change in our world daily. We have to create our own experiments that have value at a point in time to select audiences.
If you lined up everything by the patterns you’ve watched over the last 12 months, you’d likely have 12+ test plans, some repeating the same things, some not. Remember, marketing is not about finding the winner, it’s about optimizing variants for maximum yield. You have to reorient yourself to testing and tools that can help you get there.
It’s an inbox, not a billboard ad: I appreciate great creative, but as someone dedicated to sustainable outcomes, I’d rather be fast than pretty. So, when it comes to content, use what you have!
That means redesigning your content marketing plan with email in mind. It should not take a couple of days to create and proof an email. Feed it from your Web content, social content, ad content, and build templates that are adaptable and logic-driven. Your goal is the right content, not the most creative.
Unlike a billboard — which is highly visible, social-worthy and you may see multiple times in a day — email is a brief impression with a simple interaction goal. That doesn’t mean to trade the beautiful scenery for a vacation spot in the hero image with a BIG 50% off logo. It means don’t create original content for email if you don’t have to. This will increase your velocity 3X.
If something is predictable, automate it: If you can somewhat closely predict a path, flow, sequence, then automate it. That doesn’t mean you do it and leave it forever — you can still optimize it.
But you have to realize there are more customer paths than you’ll ever be able to build around, and the only way to stay close and relevant is to let machines do some of the work. If we didn’t use machines for automobile manufacturing, we’d be paying $100,000 for a car — and it’d break down every 10,000 miles.
As Oscar Wilde said, “Experience is the name everyone gives to their mistakes.” Make a few, build on this and LMK.