How to Choose What to Test (Without Overthinking It)
Lesson 1: Small experiments, big questions: What will give us the most LIFT?
If you’re bought into the idea of small experiments, the next question is: OK… but what exactly should I test?
That’s the part that trips people up.
The temptation is to test what’s easy: a subject line here, a button color there. But when your time and budget are limited, you ought to prioritize what’s impactful, not just what’s convenient.
So how do you do that?
Here’s my simple framework:
🎯 The “LIFT” Method for Picking a Smart Experiment
I ask myself:
What experiment could offer the most LIFT? (Learnings, Impact, Feasibility, and Time-to-run.)
A quick breakdown:
Learnings: Will this teach me something I can apply elsewhere?
Impact: If this works, does it meaningfully affect traffic, leads, or revenue?
Feasibility: Can I do this without wrangling 5 teams or learning a new tool?
Time-to-run: How fast can I launch and learn from this?
You don’t need a scorecard or spreadsheet (unless that’s your thing). Just gut-check your experiment idea against these four criteria. The better it performs across all four, the more likely it’s worth doing.
🧠 Examples of Smart First Experiments
Here are some solid “first tests” that often tick all four boxes:
Message clarity: Swap your headline or email subject line. Make sure it’s distinctly different from the previous version. For instance, you might want to test urgency of pain point vs. urgency of special offer. And since the average American reads at a 5th grade level, you may want to use Hemingway App to check your writing.
Channel efficiency: Compare email clickthrough vs. social CTR for the same content. Note: You may want to be extra mindful about that social CTR. The algorithms will ding your reach for embedding a link, so you’ll likely have to do some “link below in the comments” jiu jitsu.
Audience relevance: Run a SparkToro query, find several popular social accounts and popular search keywords, and run some ads to see how this audience responds to your messaging. Once you get to know this audience via SparkToro (e.g. the keywords they’re Googling, the social accounts they follow, the websites they also visit), you should have enough clues to run interest-based ads on Meta.
Offer resonance: Create a quick poll or post comparing two value props. Be sure to use your platform where you have the most reach. Which one gets more engagement?
Each of these is fast to execute, easy to measure, and could unlock bigger wins down the line.
Next week: How to design a low-lift, high-learning experiment—without spinning your wheels in prep mode.
Until then: Got a few experiments in mind but unsure which one to run first? Hit reply and tell me what you’re weighing. I’ll help you choose.
👋 See ya next Monday!