GPTrends Logo
All Posts

How are People Using ChatGPT in 2025?

Industry Trends & News
Thom Romp
Thom Romp, Product
How are People Using ChatGPT in 2025?

Yesterday Open AI plublished a fascinating in-depth growth and usage study about how peole are using ChatGPT. Here are our top take-aways 👇🏼

Growth story:

  • Up to 700M weekly active users. Not just a huge number for a (mostly) text-based product, but that's up nearly 100% from just 6 months ago and over tripled from a year ago. Having paid close attention to consumer user growth stories over the years I have to say maintaining those growth rates on such a large base is pretty crazy.
  • Messages per WAU are up at least 40% across all sign-up cohorts. Both older tenured users and new users are increasing their usage —> not just an early adopter / power user product.
  • PS: Interesting finding buried at the end: "Collis and Brynjolfsson (2025) estimate that US users would have to be paid $98 to forgo using generative AI for a month." At $20/month subscription cost is $78 per user being left on the table?

What’s it being used for?

  • Writing (the early stronghold of ChatGPT since launch) seems to be declining from 36% to 24% in the last year. I wonder how much of this is people still struggling with ChatGPT's verbose, cheesy writing style and an increased sensitivity to AI slop callouts 😬. As part of a broader trend, "asking-related" prompts are down ~25% since last year.
  • "Seeking Information" is the fastest growing category, up from 17% to 24%. More fully leveraging web search is certainly playing a big part of the story.
  • "Practical Guidance" remains the most prevalent use case: learning, how-to, health & fitness. I can personally relate, these areas feel like huge unlocks that weren't practically attainable in the Google days without deep, exhausting searching, vetting sources, avoiding ads, etc.

What we're still curious about 🤔

  • How much of the growth is due to model capability improvements versus functionality and features? The GPT-5 release felt a bit like a plateau in model performance, but ChatGPT (and OpenAI) continue to be very strong at churning out useful new features: 3rd party app integrations, memory + personalization, shopping, deep research, and agent mode.
  • Speaking of agent mode, it’s still early days, but I’d love to see their utilization and user satisfaction with the output. It will be interesting to see how we measure satisfaction and outcome success as ChatGPT (and competitors) shift from information dispensing to doing. Will time spent or messages per user drop if the application is doing more for them?