Compare
Lewsearch vs Aaru
Lewsearch publishes mean absolute error against named public polls. Aaru publishes a 0.90 median Spearman correlation from a blinded recreation of EY's 2025 Global Wealth Study. Those are different units. Aaru does not publish a pooled MAE or a price list. Choose on the rubric, not on adjectives.
What Aaru publishes
On aaru.com/simulation, Aaru states that EY asked it to recreate the 2025 Global Wealth Study, blinded: 3,600 affluent investors across more than 30 markets, finished in one day, median Spearman correlation 0.90. That is a specific, scoped result. It is not a 460-question public-poll MAE.
The same site describes a Breakwater Conviction Advantage tracker: 71 judgments across 40,000 simulated investors, dated July 17, 2026. Aaru also describes building worlds from public records (including ACS and the decennial census) plus licensed transaction, point-of-interest, search, and media data. It says it learns a network of trait relationships and samples profiles, rather than storing a finished spreadsheet of agents.
What Lewsearch publishes
7.47% mean absolute error on 404 ex-electoral questions of a 460-question benchmark vs real polls (Pew, Gallup, UT/Texas Politics Project, PPIC). Texas is 4.88% ex-electoral. California (PPIC) is 7.77%. The pre-registered held-out set, sourced after training froze, is 9.97% ex-electoral. Coverage for live panels is 27 live panels · U.S. national · 4 census regions (32 audiences). See coverage. Detail and the downloadable question file live on methodology.
Worked cell: on 18 Texas political-approval items, calibrated MAE is 3.43%. On the same Texas panel, electoral-margin items are 12.23%. Publishing both is the point. A single headline number without the hard category is incomplete.
| Rubric | Lewsearch | Aaru |
|---|---|---|
| Published accuracy unit | MAE, 7.47% on 404 questions | Median Spearman 0.90 on one EY wealth study |
| Scope of that number | Pew, Gallup, UT/TPP, PPIC, canvass; 5-fold CV | 3,600 affluent investors, 30+ markets, blinded recreation |
| Pooled public-poll MAE | 7.47% | Does not publish |
| Public pricing | Yes, on /pricing | Does not publish pricing |
| Access model | Self-serve plus demo | Enterprise / see-how-it-works |
| Other on-site figure | 178,600 simulated respondents in the published pool | Breakwater tracker: 71 judgments, 40,000 simulated investors |
Sources
FAQ
- Does Aaru publish an error rate?
- Aaru publishes a median Spearman correlation of 0.90 on a blinded recreation of EY's 2025 Global Wealth Study (3,600 affluent investors, more than 30 markets, one day). It does not publish a pooled mean absolute error against Pew, Gallup, UT/TPP, or PPIC.
- Can I convert 0.90 Spearman into Lewsearch's 7.47% MAE?
- No. Correlation and mean absolute error answer different questions. Correlation asks whether ranks move together. MAE asks how many percentage points separate option shares from a published poll. Do not translate one into the other.
- Does Aaru publish pricing?
- No. Aaru's public site does not list prices. Lewsearch publishes credit packs, a Pro plan, and an enterprise floor on the pricing page.
- What does Lewsearch publish?
- 7.47% mean absolute error on 404 ex-electoral questions of a 460-question benchmark vs real polls (Pew, Gallup, UT/Texas Politics Project, PPIC). Held-out ex-electoral MAE is 9.97%.
Check the work
Related
Canonical: https://lewsearch.com/compare/lewsearch-vs-aaru