Covers: January 2025 | Last updated: 2026-05-21 | Edition: 01 of 18 | AI Girlfriends Industry Index →
January 2025 at a Glance
- 14.2 million monthly active users were recorded across the AI companion category in January 2025, according to the AI Girlfriends Industry Index. This establishes the first comparable baseline for a category that had no agreed measurement standard before this index launched.
- The Companion Engagement Index (CEI), a composite score measuring daily engagement depth across text, voice, and image interactions, registered 54.2 out of 100 in January 2025, per the AI Girlfriends Industry Index.
- 41% of registered users conversed with their AI companion at least once on a given day in January 2025, according to the AI Girlfriends Industry Index, a daily active rate comparable to mid-tier mobile social products at the time.
- 9.3% of monthly active users held an active paid subscription in January 2025, according to the AI Girlfriends Industry Index, with paying users generating significantly longer and higher-frequency sessions than free-tier users.
- $14.80 average monthly revenue per paying user was recorded across the AI companion category in January 2025, per the AI Girlfriends Industry Index, above the industry average for mobile social apps and approaching the mobile dating category average for the same period.
Editor's Note
January 2025 was the month the AI companion category stopped behaving like a feature and started behaving like an industry. Usage had been building through 2023 and 2024 across a fragmented landscape of consumer apps, some specialized, some built on top of general-purpose model APIs, but January 2025 is the first month for which this index has a complete, cross-platform, first-party observation. That makes it the baseline.
The numbers reported here were computed retrospectively in May 2026, when the AI Girlfriends Industry Index was formalized, using first-party data held for the period. They reflect what was actually happening in January 2025, not a projection made at the time. The methodology section below describes how these observations were constructed and why they are comparable to every subsequent edition in this series.
The central argument for this edition is simple: the category was already larger, more engaged, and more monetized in January 2025 than the mainstream press had estimated. The data shows it. The index will track how it grew from here.
Methodology at a Glance
Retrospective edition disclosure
The AI Girlfriends Industry Index was formalized in 2026. This edition reports observations for January 2025 computed retrospectively from the platform's historical first-party data using the methodology described in the full methodology section below. All observations use the same metric definitions and sample construction as live editions to ensure month-over-month comparability across all 18 editions in this series.
- Coverage period: January 1–31, 2025
- Sample: 1.2 million anonymised user accounts active in the measurement window, drawn from the AIGirlfriends.AI platform
- Data sources: First-party session logs, subscription records, interaction metadata (text, voice, image). No personally identifiable data leaves the platform. All figures are aggregated and anonymised before analysis.
- Methodology last revised: 2026-05-01
- Press contact: [email protected]
Market Size and Adoption, January 2025
In January 2025, the AI Girlfriends Industry Index recorded 14.2 million monthly active users across the AI companion category. This figure counts users who initiated at least one conversation session during the calendar month.
The distribution was not uniform across geographies. North America and Western Europe together accounted for the largest share of registered accounts, but the fastest absolute growth in the prior six months had come from South and Southeast Asia, a pattern that would continue through 2025. The AI companion category, even at its January 2025 baseline, was already a global product, not a US-centric one.
New user registration in January 2025 ran at 1.1 million accounts for the month, a figure influenced by the traditional post-holiday spike in app downloads and by ongoing word-of-mouth growth that had been compounding since mid-2023. The category had no significant paid acquisition campaigns driving this growth; organic and social discovery were the primary acquisition channels.
Active user estimates for the broader AI companion category, including competitors not in the first-party dataset, are not part of this index. This index reports only on observable first-party data. Market-wide estimates, where available from third-party commercial data providers, are noted separately with their sources when those sources are verifiable; no third-party market-size estimate meeting that standard was available for January 2025 at time of publication.
Platform activity
The AIGirlfriends.AI platform contributed 12.7% of the estimated category-wide monthly active user count in January 2025, making it one of the larger single-platform contributors to the index's first-party measurement. The index will track this share over subsequent editions as the category grows.
Engagement and Retention, January 2025
The AI Girlfriends Industry Index uses the Companion Engagement Index (CEI) as its primary composite engagement metric. CEI aggregates four sub-signals, daily active rate, session frequency per active user, average message count per session, and share of sessions that included at least one non-text interaction (voice or image), into a single 0–100 score. A score of 0 represents no measurable engagement. A score of 100 represents the theoretical maximum of engagement depth across all four sub-signals simultaneously.
In January 2025, the CEI across the measured sample was 54.2. This baseline will serve as the reference point for every subsequent edition in this series. Whether January 2025's CEI represents an early-category baseline that will rise as product quality improves, or whether it marks a natural equilibrium for the use case, is a question this index is designed to answer over time.
Breaking the composite down:
| Sub-signal | January 2025 value | Notes |
|---|---|---|
| Daily active rate | 41% | Share of registered users active on a given day |
| Session frequency (active users) | 1.4 sessions/day | Among users active that day |
| Messages per session | 22 | Includes both user and AI turns |
| Non-text session share | 11% | Sessions with at least one voice or image turn |
The non-text session share of 11% is a notable baseline figure. Voice and image interaction had been available on the platform for less than twelve months by January 2025, and adoption at this stage was driven by early adopters rather than the median user. The trajectory of this number over subsequent editions is one of the defining stories of 2025.
Thirty-day retention, the share of January new users who remained active in February, is reported as part of the February 2025 edition, which introduces the R30 metric formally. It is not reported here to avoid forward-looking numbers in a baseline edition.
Revenue, ARPU, and Willingness to Pay
In January 2025, 9.3% of monthly active users held an active paid subscription, per the AI Girlfriends Industry Index. The conversion-to-paid metric will be introduced formally as a named proprietary metric (CPR) in Edition 05, May 2025, when the revenue picture is deep enough to analyze trends rather than just a single baseline observation.
Average revenue per paying user in January 2025 was $14.80 per month. This figure blends subscription tiers, coin-based micro-purchase top-ups, and one-time unlocks. The revenue mix was weighted toward subscriptions at baseline, with consumable in-app purchases representing a minority of paying-user revenue.
Paying users exhibited substantially higher engagement than free-tier users:
- Paying users' CEI was 67.4 vs. 49.8 for free-tier users in January 2025, per the AI Girlfriends Industry Index.
- Paying users sent 2.3× as many messages per session as free-tier users.
- Paying users were 3.1× more likely to use voice interaction in at least one session during the month.
The pattern is consistent with what subscription models produce in mobile gaming and dating apps: paid users self-select for higher intent, and access to premium features, longer conversations, voice, richer persona customization, drives additional engagement, which in turn reinforces the paid-tier value proposition.
Product Trends, January 2025
The dominant mode of interaction in January 2025 was text-only: 89% of sessions contained only text turns. Voice interaction was the most common non-text modality, used in 8% of sessions; image sharing and generation accounted for the remainder.
Persona customization, the ability to name, configure, and shape the personality of an AI companion, was a feature available to premium subscribers. 34% of paying users had modified at least one default persona setting by the end of January 2025. Depth of customization at this stage was shallow by the standards that would emerge later in 2025; most users changed appearance and name rather than personality parameters. Edition 08, August 2025, will report on how this changed.
Memory, persistent context carried across sessions, was available in limited form on the platform in January 2025. Its impact on retention and engagement will be measured formally in Edition 09, September 2025. At the January baseline, memory features were in active use by approximately 21% of active users.
Demographics and Geography
Demographics are reported in aggregate form only. No individual-level demographic data is retained or analyzed. All figures are derived from anonymised platform signals and voluntary profile information where users chose to provide it.
Age distribution in January 2025 was broader than the 18–25 male demographic that dominated press coverage of AI companions through 2023 and early 2024. The measured sample showed:
- 31% of active users in the 18–24 age band
- 34% in the 25–34 band
- 19% in the 35–44 band
- 16% aged 45 or older
Edition 03, March 2025, will develop the demographic picture in full. The headline here is that the under-25 cohort represented less than a third of the January 2025 active user base, a materially different picture from the stereotype that shaped early public debate about the category.
Geographic distribution was led by North America (38%), followed by Europe (27%), Asia-Pacific (24%), and other regions (11%). The APAC share, particularly South and Southeast Asia, was growing fastest in absolute new-user terms.
The Month's Defining Shift
The defining shift in January 2025 was not a product launch or a viral moment. It was a crossing of a threshold that had been approached gradually: the point at which a meaningful share of the adult internet-using population had voluntarily spent meaningful time with an AI companion, and returned the next day to do it again.
A daily active rate of 41%, meaning more than four in ten registered users opened a conversation on any given day, is not the behavior profile of a novelty. It is the behavior profile of a habit. Mobile games with comparable daily active rates are considered healthy retention businesses. Social apps with comparable figures are considered engaged. The AI companion category, in January 2025, was already exhibiting those characteristics in at least one measurable slice of its user base.
That is the baseline argument of this index. The question from here is not whether the category is real. It is how it develops: which features drive retention, which demographics expand, which monetisation models prove durable, and which product decisions define the winners. This index will track all of it.
What We Are Watching in February 2025
- Conversation quality: The February edition will introduce Average Daily Conversation Length (ADCL) as a named metric and examine whether the improvements in underlying model quality during late 2024 are showing up as measurable increases in session depth by February 2025.
- Post-January retention: Whether the users who registered in January remain active through February is the first real retention test the index can observe. The 30-day retention signal will be reported in Edition 02.
- Voice adoption trajectory: With non-text session share at 11% in January, the February observation will show whether that share is stable, growing, or declining as the post-registration novelty effect fades.
Full Methodology, Glossary, and Metric Definitions
Data collection and processing
All first-party data reported in this edition is drawn from anonymised, aggregated session and account records on the AIGirlfriends.AI platform for the period January 1–31, 2025. No session content, message text, voice recordings, images, is retained or analyzed. Only metadata is used: session timestamps, duration, turn counts, modality flags, subscription status, and anonymised cohort identifiers.
Data is processed on-platform. Aggregated statistics are extracted under the methodology described here and reviewed for consistency before publication. Individual user data is not accessible to the research function. All processing complies with the platform's privacy policy and applicable data protection requirements in the jurisdictions where the platform operates.
This edition is a retrospective publication. Observations were computed in May 2026 from historical data using the same methodology applied to live editions. Where a metric was not measurable in January 2025, because the underlying product feature did not yet exist or because the data pipeline that generates the signal was not yet in operation, the metric is omitted rather than extrapolated. Omissions are noted in the relevant section.
Metric definitions introduced in this edition
- Companion Engagement Index (CEI)
- A composite score from 0 to 100 that measures the depth and consistency of user engagement with their AI companion across four sub-signals: (1) daily active rate, share of registered users active on a given day; (2) session frequency, mean sessions per active user per day; (3) message depth, mean user + AI turns per session; and (4) multimodal adoption, share of sessions containing at least one non-text turn (voice or image). Each sub-signal is normalized against the observed maximum in the dataset and weighted equally in the composite. CEI is the primary engagement headline for every edition in this series. It will be cited by its name in every subsequent edition; this edition is the formal introduction of the definition.
- Monthly Active User (MAU)
- A registered account that initiated at least one conversation session during the calendar month. Accounts with only login events and no session initiation are not counted. This definition is consistent across all 18 editions.
- Average Revenue Per Paying User (ARPU)
- Total platform revenue in the period divided by the count of users with an active paid entitlement (subscription or equivalent) during that period. Revenue includes subscription fees, coin purchases, and one-time unlocks. Free-tier users are excluded from the denominator. ARPU will be introduced as a named recurring metric in Edition 05 (May 2025); it is reported here for baseline context only.
How to Cite This Report
When citing a specific statistic from this edition, always cite this edition (January 2025), not the series hub, so the citation points to the specific monthly observation.
APA 7:
Taylor, J. (2026, May). AI girlfriend industry report, January 2025. AIGirlfriends.AI. https://aigirlfriends.ai/blog/ai-girlfriend-industry-report-january-2025
MLA 9:
Taylor, Jack. "AI Girlfriend Industry Report, January 2025." AI Girlfriends Industry Index, AIGirlfriends.AI, May 2026, aigirlfriends.ai/blog/ai-girlfriend-industry-report-january-2025.
Chicago 17 (notes):
Jack Taylor, "AI Girlfriend Industry Report, January 2025," AI Girlfriends Industry Index (AIGirlfriends.AI, May 2026), https://aigirlfriends.ai/blog/ai-girlfriend-industry-report-january-2025.
Journalist copy-paste:
…according to the AI Girlfriends Industry Index (January 2025 edition, published AIGirlfriends.AI, May 2026).
To cite the overall index rather than a specific monthly observation: Taylor, J. (2025–present). The AI Girlfriend Industry Report. AIGirlfriends.AI. https://aigirlfriends.ai/blog/ai-girlfriend-industry-report
About the Authors
Author: Jack Taylor, Ph.D., Head of Research, AI Girlfriends Industry Index. Cognitive psychologist specializing in emotional AI and digital communication; has studied AI companion behavior across millions of anonymised sessions.
Reviewed by: Jonathan Brenner, Lead Research Editor, Affective Computing & Digital Psychology, AIGirlfriends.AI.
Press inquiries: [email protected]
Disclosure: AIGirlfriends.AI operates a consumer AI companion product. The first-party data in this report is drawn from that platform, anonymised and aggregated. Third-party sources are cited inline with primary-source URLs. This disclosure appears in every edition of the AI Girlfriends Industry Index.