Retention is the the only test that truly matters

It tells you whether users come back after the novelty fades.

This App Store ranking chart tracks major AI apps over 3 years.

TL;DR?

- ChatGPT is defining the category’s retention benchmark.

- Perplexity proves consumers will swap “10 blue links” for answer‑first search, using trust-through-transparency as the differentiator.

- Brand clarity, not feature breadth, is the new growth hack. Every spike‑and‑crash pattern on the chart is a branding, access or UX tax in disguise.

You need to be one long‑press, hot‑key, or “@perplexity” away whenever the user thinks, “I need help”.

For builders & investors

1. App‑store rank ≠ model IQ 

2. Feedback loops and UX polish decide the leaderboard.

3. 'Re‑imagined search' is wide open. Perplexity’s rise is proof of white space. But Google has tremendous leverage in distribution (chrome, gmail, android, maps) and can use it to dominate. 

4. Research leadership doesn’t guarantee UX leadership (e.g. Gemini).

5. Platform extensibility (GPTs, plugins, memory) compounds faster than raw params.

6. When you already own distribution, even a 1.0 product can blitz the charts. Design your GTM accordingly.

Microsoft proves default beats discovery. 

Copilot arrived late yet rocketed up the charts on the back of Office365 placement. 

Distribution still eats product when you sit on a billion desktops.

Other signals that can be inferred.

- ChatGPT’s uninterrupted blue ceiling.

 Winner‑take‑most dynamics are already here. 

 Dethroning requires a 10x better product, not incremental features.

 Which is why OpenAI can get away with one of the most confusing product naming systems in the tech industry!

- Microsoft Copilot’s pogo‑stick spikes. 

 Each orange surge is followed by a crash below #100. 

 Users try it, churn, and Microsoft relaunches. 

 Is naming pivots from “Bing → Copilot → Copilot Pro” creating branding debt that shows up as ranking volatility?

 Useful to note that Microsoft has mastered the distribution-before-product play. O365 bundling means even an unremarkable Copilot app can scale instantly.

- Google Gemini’s mid‑table slide

 Google is defending the search bar, not the App Store.

 An API‑first, mobile‑second posture limits consumer mindshare. 

- Perplexity’s 0‑to‑top‑100 sprint 

 A conversation‑first search UX plus citations beats a decade of link‑based habit. 

 Trust‑through‑transparency is monetisable.

- Claude AI’s ghost trace

 Claude is playing a low-profile, high-performance B2B game. 

 Anthropic is staying API-only atm and betting on becoming AI infrastructure, not a consumer brand. 

 B2B revenue is great, but skipping consumer distribution forfeits the feedback loop that hardens PMF.

- No one crosses the ChatGPT floor

 Looks like the TAM is winner‑take‑most until a new modality (e.g. OS‑level agents) resets the playing field.

- Flat vs. saw‑tooth = Retention proxy

 Stable lines imply DAU/MAU of atleast 0.4. 

 Erratic lines indicate novelty churn.

- The cost of modal overload

 Single‑job apps win on mainstream retention. 

 Can breadth wait until habit forms?