The science behind the score.
Every ViewShift Lift result is proven in a randomized controlled trial, the same rigorous testing environment used in medical research. Here's exactly how it works, start to finish.
The method, end to end
How a ViewShift Lift test works.
Scroll through the experiment: one stage at a time, from a single panel of real people all the way to a weighted, statistically grounded score.
The problem
Did your content persuade them, or did they simply engage with it?
Your content can rack up views, clicks, and shares and still not change a single mind. Engagement tells you people noticed; it can't tell you they were persuaded. And most measurement only sees what happened after your content went out, never what would have happened anyway. A rise in favorability might be your content working, or it might be a trend you happened to ride. Even an A/B test can't fully settle it: the people who saw version A aren't the same as those who saw version B. To know your content caused the change, you need a fair comparison: two groups that are alike in every way except one, whether or not they saw it.
Step 1: The sample
It starts with a large, representative panel.
Every study begins by recruiting real people who reflect the audience you care about. We set quotas on age, gender, race, and ethnicity so the right kinds of people show up in the right proportions, then balance the group against the real population. The result is a sample that mirrors who you're trying to reach, not just who was easiest to survey.
Step 2: Audience & quality
Then we tune it to your exact audience.
We target the exact audience your decision depends on, then confirm we can field it at a sample size large enough to be statistically sound. Every response is held to ViewShift's quality-control standards, so the data behind your score is reliable, representative, and ready to defend, on every study.
Step 3: Random assignment
Next, we split them at random.
This is the move that makes everything causal. The moment someone enters the study, they're randomly assigned to see your content or a neutral baseline. The assignment is completely random. It ignores who they are and what anyone before them saw, so the two groups start out alike on average. Any difference we measure later can come from only one thing: what they saw.
Step 4: One difference
Two groups. One difference.
The control group sees a neutral baseline. The test group sees your content: a video ad, a social or display ad, a tagline, a value proposition. Everything else about their experience is identical. That single controlled difference is the heart of a randomized controlled trial, the same design used to test medicines, applied to your content.
Step 5: The same questions
Both groups answer the same survey.
Right after exposure, every respondent answers the same set of questions: the outcomes you care about, like whether they'd consider you, trust you, or take action. Because exposure happens inside the study, nearly everyone who starts finishes, so we're not left guessing about the people who clicked away before the question arrived.
Step 6: Measuring the response
We then capture how each group responded.
Now there are two readings of the very same questions: how the baseline group answered, and how the group that saw your content answered. Set side by side, they reveal exactly what your content changed, the clean comparison that traditional before-and-after tracking can only approximate.
Step 7: Weighting to the real world
And we weight the results to match the population.
No sample is ever a perfect mirror of the population, so we correct for it. Using a technique called raking, we give each respondent's answers slightly more or less influence until the group matches the real audience on age, gender, race, education, and party, plus recent presidential vote for broad audiences. Extreme weights are trimmed so no single person can swing the result. It's how the score stays honest and projectable to the people you want to move.
Step 8: The Lift Score
Your Lift Score shows how your message performed against the baseline.
The intuition is simple: the score is the gap between the test group and the baseline. Under the hood we estimate that gap with a statistical model that adjusts for any leftover differences between the groups (age, gender, race, party, education, and past vote), so the number reflects your content, not a fluke of who landed in which group. We report it with a margin of error, the range the true effect almost certainly falls within, and when chance can't plausibly explain the gap, we mark the result statistically significant. Every piece of content lands somewhere on a clear scale from negative lift to positive lift.
Step 9: Every audience, not just the average
Finally, we read it for every subgroup.
A single average can hide the story. Using the same model, we re-read the result inside every segment that matters, with subgroups tailored to your industry, category, and brand: demographics, regions, customer tiers, buyer types, however your market divides. We surface the audiences where your content clearly moved people, or clearly backfired. That's usually where the next decision lives.
Proof in days, not months
Rigor you can act on.
This is the gold standard of causal measurement, run fast enough to act on. No selection bias dressed up as insight, just a clear, weighted, statistically grounded answer to the only question that matters: did it work?
See the method on your own content
Run the experiment on your content.
The fastest way to understand the method is to watch it run on something real. Bring a piece of creative or a draft statement and we'll show you the score, the confidence interval, and the subgroups behind it.