Research

How We Validate the IMPRS Score

Inside the V32 study: 122 companies, 551 observations, and score movement that tracked next-year revenue movement at a rank correlation of +0.395.

Amilere Research Dec 2025 12 min read

The Central Question

Does brand health track the revenue that comes later? That is the only question that matters for a score like ours. So we test every formula release against subsequent revenue outcomes. The current release is V32, validated in July 2026. The study covered 122 companies, 551 clean observations, and 673 pipeline runs.

Movement Is the Signal

The strongest test is movement. When a company's score moved, did revenue move the same way, and by how much? The size of the score change tracked the size of the next-year revenue change at a rank correlation of +0.395 across all segments, and +0.563 for consumer brands. Raw direction-match on its own is a weak test, because revenue rises in most years anyway, so we report the correlation and the ordering instead. We do not claim the score forecasts revenue perfectly. No honest metric can.

Outcome Ordering

We also test ranking. If company A outscores company B, does A see the better revenue outcome the next year? Across the full dataset, growth ordering matched 56.5% of the time, rising to as high as 86.6% when the score clearly separated two companies. For Fortune 500 brands, the score level correlated with next-year growth at r +0.386. For creators, the score level is the strongest single signal in the panel at +0.476, though creators should be compared within their own tier.

Why Lag Matters

Marketing works on a delay. Content you publish today builds equity that converts later. So we test score movements against later revenue, not just same-period revenue. This lag-aware design keeps the test honest. It also matches how you should read your own score: as a leading signal, not a same-day receipt.

What This Means for Your Score

The formula is versioned, and V32 is current. Every release must pass validation before it ships. The dataset grows, and the formula is recalibrated against it. Your score from 350 to 900 rests on this process. Read it as a tested signal: better than chance, and honest about how much better.

Key Takeaways

  • 1V32 validation (July 2026): 122 companies, 551 observations, 673 pipeline runs
  • 2Score movement tracked next-year revenue movement at rank correlation +0.395, stronger than the score level alone
  • 3Growth ordering: 56.5% overall, up to 86.6% when the score clearly separates two companies
  • 4Fortune 500 score level correlated with next-year growth at r +0.386
  • 5Every formula release is validated against revenue before it ships