ChatGPT's Users Just Went Global. Your Audience Isn't Who You Think.
On 30 June 2026, OpenAI published its Signals data on how ChatGPT adoption has expanded, and buried inside the charts is a finding most marketers have not absorbed: users predominantly using a language other than English now represent over half of active users, and usage by people with typically-female names now makes up most usage globally. The centre of gravity for AI-assisted discovery has moved. If your GEO strategy still assumes an English-speaking, male-skewed, North American core, you are optimising for a shrinking slice.
Let me be plain about what this is. OpenAI Signals measures aggregated usage across Individual plans (Free, Go, Plus and Pro). It is the clearest public read we have on who is actually using ChatGPT and how that behaviour deepens over time. And the story it tells is one of breadth, not just scale.
What the data actually shows
Three shifts stand out, and each one has a practical consequence.
1. People do more, and do more things, the longer they stay
Six months after signing up, users sent 50% more messages per day than they did when they joined. Over the same period they doubled the number of distinct tasks they tried. OpenAI splits this into two measures: depth (messages in a trailing 28 days) and breadth (cumulative capabilities used, across 53 task categories).
From my observation, this is the part growth leaders keep missing. ChatGPT is not a one-trick search box. As users mature, they fold it into more of their routine, and each new task category is a new place your brand can either be surfaced or be absent.
2. The fastest growth is not where you are spending
Since July 2023, adoption has grown across every continent, but the fastest relative growth has been in Africa and Asia. The same pattern holds by development level: lower-Human Development Index (HDI) countries have seen the fastest relative growth in weekly active users, helped by low-cost Free and Go plans.
This could reshape how you think about total addressable market. The next wave of AI-native buyers, students and professionals is not concentrated in the markets where digital ad budgets have historically pooled.
3. The user base is more female and less English than the cliche suggests
Usage by people with typically-female names has increased and now represents most usage globally. Brazil, Colombia, Poland and Namibia rank among the countries where messages from users with typically feminine names most exceed those with masculine names. On language, Spanish, Portuguese and Arabic lead the non-English pack, while Uzbek, Kazakh and Burmese posted the largest percentage increase in share since July 2023.
The shift in numbers
| Signal | What OpenAI reported | Why it matters |
|---|---|---|
| Depth (6 months in) | 50% more messages per day vs signup | Engagement compounds; recurring surface for discovery |
| Breadth (6 months in) | Distinct tasks tried doubled | More task categories = more entry points |
| Fastest regional growth | Africa and Asia | New demand outside legacy ad markets |
| Language mix | Non-English now over half of active users | English-only content is a minority strategy |
| Leading non-English languages | Spanish, Portuguese, Arabic | Clear localisation priorities |
| Fastest-rising languages | Uzbek, Kazakh, Burmese | Emerging markets to watch |
Source: OpenAI Signals, data through 31 May 2026.
How this changes what you do
The methodology is worth noting before you act on it. Depth and breadth come from a 0.1% sample of accounts created between 15 October 2025 and 1 May 2026, with activity tracked through 31 May 2026. Gender is estimated from name-to-gender crosswalks, not collected directly, so treat it as a directional signal rather than a headcount. Even with those caveats, the direction is unambiguous.
Here is what I would prioritise:
- Audit your content for language coverage. If more than half of active users are working in a language other than English, English-only assets cannot be your whole GEO play. Start with Spanish, Portuguese and Arabic.
- Map your brand against task categories, not keywords. With users doubling the tasks they try, ask where your product could plausibly appear across research, planning, drafting and comparison prompts.
- Rethink your market map. Fast relative growth in Africa, Asia and lower-HDI countries is a demand signal worth testing before competitors treat it as obvious.
- Design for the actual user. A more female, more global base should shape tone, examples and imagery, not just translation.
I think the wider lesson connects to how discovery itself is fragmenting. As I argued in my piece on GPT-Live turning voice AI into a discovery channel, the question is no longer just where people search but how and in which language. And the case for consolidating your inputs, which I covered in building a stronger marketing data foundation, only gets sharper when your audience is this distributed.
The concrete action this week: pull your own analytics by language and region, compare it against the Signals trend lines, and identify the one non-English market where you are already getting traffic but publishing nothing. That gap is your fastest win. Researchers can also download the OpenAI Signals dataset to model it against their own numbers.
Tags