Precision Preview is live in GeoLift by Recast, and it’s already changing how teams think about designing lift tests.
Here’s the problem it solves: when teams design geographic-based incrementality tests today, they often size them using power analysis – a statistical exercise that answers the question “will I detect a statistically significant effect?” That might be a fine question for an academic paper, but it’s not the right question for a marketing team trying to make budget decisions.
For marketers, an experiment they run with 80% power can still come back with an incremental ROI estimate that ranges from 0.5x to 10x. That result is technically significant, but it’s also practically useless.
Nobody is going to reallocate a $5M channel budget based on an experiment result that’s “somewhere between 0.5x and 10x.” The question that actually matters for these teams is: will the results be precise enough for my team to act on?
That’s what Precision Preview answers.
What it does
When you design an experiment in GeoLift by Recast using the REBA algorithm, Precision Preview shows you the expected ROI or CPA confidence intervals at different test investment levels – before you actually commit a dollar to the test.
So at low test spend levels, a well-designed experiment produces a massive ROI range because the experiment doesn’t have enough signal to give a very precise answer about the true lift.
Precision Preview makes this relationship visible so your team can determine the appropriate investment level where the resulting range is useful to make decisions with.
How teams are using Precision Preview
There are three primary use cases for this new GeoLift by Recast feature.
- Right-sizing experiment spend. Finding the minimum investment that produces a confidence interval narrow enough for the decision your team has at hand.
- Killing bad tests before they waste money. If the expected precision is too wide for your team at any feasible budget, you know this before committing any spend.
- Getting test designs pre-approved by finance. With Precision Preview, you can now say “this experiment will produce an ROI estimate of 3x ± 0.6x”. That’s a sentence a CFO can evaluate in a way that “we have 80% power” is not.
How to access it
Precision Preview is available now for all GeoLift by Recast users. You’ll find it at the bottom of the deep dive power analysis when you design an experiment using the REBA algorithm. If you’re new to GeoLift by Recast, you can start a trial that’s completely free for six months.
We’re shipping a lot of updates to GeoLift by Recast and across Recast’s platforms. This is one we think will meaningfully change how teams plan experiments, and we’d love to hear how you use it.
