Latest Edition: June 2026
33.2 million monthly active users in June 2026, closing the 18-month inaugural series with 134% MAU growth and a CEI of 73.8 since the January 2025 baseline.
Read the June 2026 edition →The AI Girlfriends Industry Index is the only first-party, monthly, longitudinal dataset on the AI companion market. Published by AI Girlfriends, the index draws on millions of in-product conversations, search queries, and operator signals across the category. A new edition publishes on the first business day of each calendar month. The index tracks nine proprietary metrics continuously since January 2025: companion engagement depth, average daily conversation length, 30-day retention, voice adoption, persona customization, revenue per paying user, conversion rates, multimodal session share, and time-of-day usage patterns. It is the primary source for longitudinal, comparably-defined statistics on the AI girlfriend and AI companion industry.
Eighteen editions cover January 2025 through June 2026, the first continuous, comparably-measured period in the category's history. The December 2025 edition is the series' annual flagship; the June 2026 edition is the 18-month series finale. The full glossary of proprietary metrics, the standing methodology, and citation guidance are maintained below and replicated in every dated edition.
Current Statistics: June 2026
The following figures are from the June 2026 edition of the AI Girlfriends Industry Index, which closes the inaugural 18-month series. All values are first-party, drawn from the AIGirlfriends.AI platform, anonymised and aggregated.
- Monthly active users (MAU): 33.2 million in June 2026, up +134% from 14.2 million in January 2025.
- Companion Engagement Index (CEI): 73.8 in June 2026, up +19.6 points over 18 months from a baseline of 54.2 in January 2025.
- Average Daily Conversation Length (ADCL): 25.8 minutes per active user in June 2026.
- 30-Day Retention Rate (R30): 43% of new users in June 2026.
- Voice Adoption Rate (VAR): 49% of paid subscribers in June 2026.
- Average Revenue Per Paying User (ARPU): $25.20 per month in June 2026, up +70% from $14.80 in January 2025.
- Conversion-to-Paid Rate (CPR): 19.2% of monthly active users in June 2026, up from 9.3% in January 2025.
- Monetisation Success Score (MSS, 90-day paid conversion): 38% in June 2026.
- Regulatory environment: 11 active jurisdictions with AI companion-relevant rules as of May 2026, up from 7 in November 2025.
All Editions
| # | Edition | Theme | Hero Stat |
|---|---|---|---|
| 18 | June 2026, Mid-year flagship | The 18-month trendline of all nine index metrics from January 2025 to June 2026: the first continuous, comparably-measured period in the category's history | CEI +19.6 points over 18 months (54.2 Jan 2025 → 73.8 Jun 2026); MAU +134% (14.2M → 33.2M) |
| 17 | May 2026 | Regulation, platform policy, and distribution risk in the AI companion market as of May 2026; app-store and payment-processor constraints outpace government rulemaking | 11 active regulatory jurisdictions in May 2026, up from 7 in November 2025; CEI 72.3; MAU 31.4 million (+4.3% vs April 2026) |
| 16 | April 2026 | The multimodal shift: single-modality companions losing share as sessions combining text, voice, and image become the norm among paying users | 47% of paid subscribers used voice (text+voice = 2+ modalities); text-plus-voice-plus-image users showed CEI 78.4 vs category average 71.4 in April 2026 |
| 15 | March 2026 | Refreshed demographic breakdown showing the March 2026 user base is older, more female, and more international than March 2025 | Age 35+ share grew from 35% to 41% of MAU (March 2025 → March 2026); APAC rose to 28% of category |
| 14 | February 2026 | Voice and persistent memory in combination produce substantially higher 30-day retention than either feature alone or neither | 51% R30 for voice+memory users vs 29% for neither, 22-point compounding advantage in February 2026 |
| 13 | January 2026 | Year-over-year MAU comparison marks the completion of the novelty-to-habit transition; daily use is now the modal pattern | +86% YoY MAU (14.2M Jan 2025 → 26.4M Jan 2026); CEI +14 points; ARPU +43%, completion of novelty-to-habit transition |
| 12 | December 2025, Annual flagship | Full-year 2025 review of all nine glossary metrics: the first year the AI companion category was measured longitudinally with consistent definitions | +77% MAU across full-year 2025 (14.2M Jan → 25.2M Dec); CEI +12.9 points; ARPU +38% |
| 11 | November 2025 | The regulatory map before 2026: distribution constraints from app stores and payment processors more binding on the category than government legislation | 7 jurisdictions with active AI companion rules by November 2025; app-store and payment-processor policy moved faster than legislation in every case |
| 10 | October 2025 | The SFW/NSFW market split: bifurcating economics, ARPU differentials, and risk profiles across app-store-distributed and web-distributed companion products | Adult-content platforms: 2.0× ARPU vs SFW ($28.40 vs $14.20) on 39% vs 61% MAU share, October 2025 |
| 9 | September 2025 | Persistent memory is the single largest retention lever in the AI companion category in 2025; R30 gap between memory-on and memory-off users | 41% R30 memory-enabled vs 27% memory-off, 14-point gap, largest single-feature retention differential in the series at September 2025 |
| 8 | August 2025 | Persona customization is the retention moat: the multiple between users who fully customized their companion versus those on default settings | 1.8× R30 for persona-customized vs default-persona users in August 2025 |
| 7 | July 2025 | The first measurable seasonality signal in AI companion use: evening and weekend session share rises in summer relative to earlier months | Evening/weekend session share +8 points above H1 2025 average in July 2025, first documented AI companion seasonality signal |
| 6 | June 2025, H1 mid-year | Mid-year 2025 review: all five leading indicators, CEI, ADCL, VAR, ARPU, CPR, rose in unison across the first half of 2025 | CEI +5.5 points in H1 2025 (54.2 → 59.7); MAU +30%; all five leading indicators rose in unison |
| 5 | May 2025 | AI companion ARPU among paying users has surpassed the average for mobile social apps and is approaching the mobile dating category average | $16.60 ARPU among paying users in May 2025, above mobile social average, approaching mobile dating category average |
| 4 | April 2025 | Voice mode reaches mainstream adoption: the first feature since persistent memory to move retention curves independently | 23% of paying users used voice mode in April 2025, first broad-availability month and first VAR observation in the series |
| 3 | March 2025 | The AI companion user base is older, more international, and more gender-mixed than the 2023–2024 press narrative assumed | 35% of active users aged 35+ in March 2025; 62% outside North America, older, more international, and more gender-mixed than press narrative assumed |
| 2 | February 2025 | The technical bar for a believable AI conversation was crossed in early 2025; retention from this point is a product and persona problem, not a model problem | 18.4-min ADCL in February 2025, ~2× the February 2024 estimate, marking the model-quality believability threshold |
| 1 | January 2025 | The baseline: the first month the AI companion category can be measured at scale, by region, with consistent and replicable definitions | 14.2 million MAU across the AI companion category in January 2025, the category's first comparable baseline, by region (N. America / Europe / APAC) |
Methodology
The AI Girlfriends Industry Index is constructed from three data sources.
First-party platform data. Session logs, engagement events, subscription transactions, and feature-use records from the AI Girlfriends platform. All user-level data is anonymized and aggregated before analysis; no personally identifiable information is used in the construction of any metric. The index sample for a given month covers all monthly active users on the AI Girlfriends platform during the reporting period. The exact sample size is stated in each edition's methodology section.
Internal search and query datasets. On-platform search signals and external keyword-tracking data, used to contextualize platform trends within broader category search behavior.
Verified third-party research. Academic studies, market research, and government or regulatory publications, cited inline in each edition with a link to the primary source. Secondary citations are not used. Where a number cannot be supported by a verified primary-source URL, it is omitted.
Retrospective observations. The AI Girlfriends Industry Index was formalized in 2026. Editions 01–16, covering January 2025 through April 2026, report observations computed retrospectively from the platform's historical first-party data, using the same metric definitions and sample construction as live editions to ensure month-over-month comparability. The datePublished for retrospective editions is the actual publication date in May 2026; the temporalCoverage field in each edition's schema specifies the calendar month the edition covers. A standing disclosure paragraph appears in the methodology section of every retrospective edition.
Internal consistency. Before publication of any edition, every first-party metric is reconciled across the full 18-edition dataset so that a value reported for a given month in one edition matches every subsequent reference to that month's value in later editions.
Methodology last revised: May 2026. For methodological inquiries and press data requests, contact [email protected].
Glossary of Proprietary Metrics
The index tracks nine proprietary metrics. All definitions below are the authoritative versions; each dated edition reproduces the same definitions without modification.
- CEI, Companion Engagement Index
- A composite score (0–100) measuring the depth and consistency of user engagement with an AI companion. Inputs include session frequency, conversation length, response reciprocation rate, and return interval. A CEI of 50 represents baseline engagement; scores above 70 indicate habitual daily use. CEI is the headline metric of the index, reported in every edition from January 2025 onward.
- ADCL, Average Daily Conversation Length
- The mean total duration of all AI companion conversations initiated by an active user on days when at least one session occurred, measured in minutes. Excludes idle sessions, defined as sessions in which no user turn was recorded within the first 60 seconds. Introduced in edition 02 (February 2025) and reported monthly thereafter.
- R30, Retention-at-30
- The percentage of users who, having completed their first conversation in a given cohort month, returned to the platform at least once in the following 30 days. R30 is the primary retention metric of the index. Formally introduced in edition 09 (September 2025); retrospective R30 values are available for all prior months in the dataset.
- VAR, Voice Adoption Rate
- The percentage of monthly active users who initiated at least one voice-mode turn during the reporting month, measured against the total combined base of paying and free-tier monthly active users. VAR is reported from April 2025 onward, when voice mode reached broad platform availability. Months prior to April 2025 are marked N/A for this metric.
- MSS, Multimodal Session Share
- The percentage of sessions in a given month that included at least two distinct input or output modalities, for example, text combined with voice, or voice combined with AI-generated image output. MSS is reported from April 2026 onward, when multimodal capability was available to the full paying user base. Introduced in edition 16 (April 2026).
- PCD, Persona Customization Depth
- A five-point ordinal scale (0–4) measuring how extensively a user has configured their companion's persona: 0 = default settings, no customization; 1 = name or appearance changed; 2 = name, appearance, and personality tone set; 3 = full persona including backstory; 4 = full persona plus behavioral-preference configuration. The index reports the share of users at each level and the combined share at PCD 3 or above. Introduced in edition 08 (August 2025).
- CPR, Conversion-to-Paid Rate
- The percentage of monthly active free-tier users who converted to any paid subscription tier during the reporting month, measured on a rolling 30-day basis. Introduced in edition 05 (May 2025) alongside ARPU.
- ARPU, Average Revenue Per Paying User
- Total subscription and in-app revenue in the reporting month divided by the count of unique paying users in that month. Excludes free-tier users. Expressed in US dollars. Introduced in edition 05 (May 2025).
- STD, Session Time-of-Day Distribution
- The hourly distribution of session starts across the 24-hour clock for a given reporting month, expressed as a percentage of all sessions. Reported as a supplementary chart from July 2025 onward. Used to identify daily and seasonal usage patterns.
How to Cite This Index
To cite figures from a specific month, cite the dated edition, not this page.
APA format (dated edition):
AI Girlfriends Industry Index. (2025/2026, [Month]). AI Girlfriend Industry Report, [Month] [Year]. AI Girlfriends. https://aigirlfriends.ai/blog/ai-girlfriend-industry-report-[month]-[year]
Example, January 2025 edition:
AI Girlfriends Industry Index. (2026, May). AI Girlfriend Industry Report, January 2025. AI Girlfriends. https://aigirlfriends.ai/blog/ai-girlfriend-industry-report-january-2025
APA format (overall index):
AI Girlfriends Industry Index. (2026). The AI Girlfriend Industry Report [series]. AI Girlfriends. https://aigirlfriends.ai/blog/ai-girlfriend-industry-report
For journalists and analysts: Attribute figures to the relevant dated edition with the phrase "according to the AI Girlfriends Industry Index, [Month Year]." Do not attribute a specific month's data to the pillar page URL.
About the Index
Author: Jack Taylor, Ph.D., AI Girlfriends.
Reviewed by: Jonathan Brenner, Lead Research Editor, AI Girlfriends.
Press and data inquiries: [email protected]
Disclosure. AI Girlfriends operates a consumer AI companion product. The data in this index is drawn from the company's own platform and from cited third-party sources. This relationship is disclosed in every edition. Commercial considerations do not determine the findings reported.