Search vs. Store — AI app search demand vs App Store rank vs web traffic (US, 2026)

One row per major AI app joining three independent popularity signals. Two cross-sectional US snapshots (2026-07-01 and 2026-07-26), 40 curated AI apps each. Aggregate/derived only — no raw records, no PII. License: CC BY 4.0.

• Study + methodology: https://crawlora.net/blog/search-vs-store-2026?utm_source=github&utm_medium=referral&utm_campaign=search-vs-store
• Dataset page: https://crawlora.net/datasets/search-vs-store?utm_source=github&utm_medium=referral&utm_campaign=search-vs-store
• Living index (daily panel + lead/lag test): https://crawlora.net/search-vs-store?utm_source=github&utm_medium=referral&utm_campaign=search-vs-store

The finding

Rank the 40 apps by web visits and you get roughly the search order (ChatGPT, Gemini, Canva, Claude, DeepSeek…), not the App Store order. Search demand and web traffic agree; the App Store rank is the outlier, scrambled by distribution deals and platform push. The divergence column makes that explicit per app.

Columns

• app — display name.
• trends_term — the exact query measured on Google Trends (hand-picked to disambiguate brand names, e.g. "Claude AI", "Google Gemini").
• search_tier — coarse Google search-demand tier: Dominant > High > Moderate > Low > Minimal. Coarse on purpose (see caveats).
• trends_interest_chatgpt100 — trailing-12-month mean Google Trends interest, anchor-normalized so ChatGPT = 100. Directional, not precise for the small tail.
• best_app_store_rank — best current US chart position across iOS/Play, Top Free + Grossing (null = off the top-100). ios_rank / android_rank — per-store best.
• chart_score — 101 - best_app_store_rank (higher = better standing; null if off-chart).
• ratings_count — combined iOS+Play ratings count (install proxy), where matched.
• sw_monthly_visits — SimilarWeb estimated monthly web visits (absolute; blank = below SimilarWeb's floor). sw_global_rank — SimilarWeb global site rank.
• divergence — downloads_not_search | search_not_downloads | aligned.

Files

File	What
data/search-vs-store-ai-apps-2026-07-26.csv	Snapshot 2026-07-26, one row per app
data/search-vs-store-ai-apps-2026-07-26.json	Same rows + metadata + the search x store tier matrix
data/search-vs-store-ai-apps-2026-07-01.csv	Snapshot 2026-07-01, one row per app
data/search-vs-store-ai-apps-2026-07-01.json	Same rows + metadata + tier matrix

The .json files are a metadata wrapper (method, columns, tier_matrix, data[]), so the CSVs are the flat, directly-loadable tables.

Caveats worth reading before you cite this

• Do not compute a cross-app search correlation from trends_interest_chatgpt100. Google Trends rescales 0-100 against the largest term in each request, and ChatGPT out-searches the field ~30-1000x, which collapses the tail into noise. The tiers and the divergence label are robust; a correlation coefficient over these values is an artifact of lost precision.
• sw_monthly_visits / sw_global_rank are joined from the nearest available panel date when the daily SimilarWeb panel had no row for the snapshot date (see similarweb_as_of in the JSON metadata) — a few days off from the rest of the row.
• These are point-in-time snapshots, not a long rank history: upstream App Store chart history is only ~2-3 weeks deep.

Mirrors

Where	Link
Hugging Face	https://huggingface.co/datasets/crawlora-net/search-vs-store
Kaggle	https://www.kaggle.com/datasets/plantpark/search-vs-store
Zenodo (DOI)	https://doi.org/10.5281/zenodo.21987676

Citation

Crawlora (2026). Search vs. Store: AI App Search Demand vs App Store Rank vs Web Traffic (US, 2026). CC BY 4.0. https://doi.org/10.5281/zenodo.21987676

Collected with Crawlora.