Dating Trust Index 2026: Global Suspicion Search Trends
Original anonymized data from 14,000+ relationship trust searches in 136 countries: when suspicion peaks, where it comes from, and the AI referral shift.
Headline findings
- Interest in relationship trust searches more than tripled in July 2026 and stayed at that level through August.
- Suspicion peaks between 12:00 and 17:00 UTC — lunchtime and early evening — and bottoms out at 05:00 UTC, a 3.4x difference.
- 61% of searches happen on a phone.
- AI assistants (ChatGPT and Perplexity) now refer about 14% of all human visits — roughly 89 times more than Bing.
About this index
PartnerCheck is a photo-based check of public environmental signals: a user uploads one photo and receives an indicative summary of matching public material. Since February 2026 we have recorded anonymous, aggregate usage statistics about how people find and use the service. This page publishes those aggregates — and only those aggregates — as a citable dataset.
The sample is 14,423 recorded visits between February 24 and August 22, 2026, including 962 completed photo checks. Nothing on this page can identify a person: no names, emails, IP addresses, session identifiers, cities, devices fingerprints or search subjects exist in this dataset, and any group smaller than 50 records is suppressed entirely. The full methodology, data dictionary and a free download live on the methodology page.
1. The 2026 suspicion surge
Monthly visit volume was flat through the spring — roughly 1,200 visits a month from March to June. Then July arrived. Volume jumped to 4,240 visits, 3.6 times the June figure, and August is running ahead of July on a per-day basis (about 200 visits per day versus 137).
Recorded visits per month, 2026. *Partial months: tracking began February 24; August figures cover August 1 to 22.
The same shift shows up in completed checks: 689 of the 962 photo checks ever run on the service — 72% — happened after July 1. We cannot say from this data why interest surged. What we can say is that the surge coincides with a structural change in where visitors come from, covered in section 5.
2. When suspicion peaks
Aggregating all human visits by hour (UTC) produces a remarkably consistent shape. Activity climbs through the morning, peaks in a broad window from 12:00 to 17:00 UTC — lunchtime and early evening in Europe, morning to midday in the Americas — then fades overnight. The single busiest hour, 14:00 UTC, sees 892 visits. The quietest, 05:00 UTC, sees 260. Peak to trough, that is a 3.4x difference.
Human visits by hour of day (UTC), February 24 to August 22, 2026. Automated bot traffic excluded.
Read plainly: people look for answers about a partner when the day gives them a private moment — the lunch break, the commute home, the early evening — not in the middle of the night. The 05:00 UTC trough is the only hour where the world is, statistically speaking, at peace.
3. The device of suspicion is the phone
Of 12,290 human visits with a resolved device class, 7,492 (61%) came from a mobile phone, 4,783 (39%) from a desktop or laptop, and 15 (0.1%) from a tablet. Suspicion is a mobile behavior: it happens on the device people carry into the hallway, the bathroom, the parked car.
4. Where the searches come from
A country was resolved for about two thirds of human visits (early tracking did not record geography, so roughly a third of visits are unassigned). Among geo-resolved visits, the United States dominates, followed by Italy — a share far out of proportion to Italy's population, which suggests the topic resonates strongly in the Italian-language market — then Germany, the United Kingdom and India.
| Country | Visits | Share of geo-resolved |
|---|---|---|
| United States | 2,272 | 27.7% |
| Italy | 1,523 | 18.6% |
| Germany | 420 | 5.1% |
| United Kingdom | 396 | 4.8% |
| India | 322 | 3.9% |
| Australia | 267 | 3.3% |
| Netherlands | 239 | 2.9% |
| Canada | 223 | 2.7% |
| Czechia | 211 | 2.6% |
| France | 178 | 2.2% |
Human geo-resolved visits, February 24 to August 22, 2026 (n = 8,210; bot traffic excluded). In total, visits were recorded from 136 countries; countries below the minimum cell size of 50 are not listed.
5. The AI referral shift
The most important structural change in this dataset is how people arrive. Of 12,725 human visits with a classifiable referrer, Google sent 3,122 (24.5%) — but ChatGPT sent 1,759 (13.8%). Add Perplexity and AI assistants account for about 14% of all human traffic. Bing, the world's second-largest traditional search engine, sent 20 visits (0.2%). In this niche, AI assistants refer roughly 89 times more visitors than Bing.
| Referral source | Visits | Share of human traffic |
|---|---|---|
| 3,122 | 24.5% | |
| Direct / unknown | 2,934 | 23.1% |
| ChatGPT | 1,759 | 13.8% |
| TikTok | 520 | 4.1% |
| Bing | 20 | 0.2% |
| Perplexity | 18 | 0.1% |
| All other referrals | 4,352 | 34.2% |
Human visits by referrer class (n = 12,725), February 24 to August 22, 2026. Direct includes visits where the referrer was stripped by a browser, app or payment redirect.
The mechanism matters for anyone publishing in sensitive verticals. AI assistants do not rank ten blue links — they cite one or two sources inside a generated answer. In this dataset the citations cluster on practical, reference-style guides rather than product pages, which is consistent with how these systems choose sources elsewhere.
6. What people read before they search
The most-visited content page is the guide on why reverse image search fails on dating apps (631 human visits), followed by the free reverse image search comparison (423) and the Tinder profile search guide (176). Reference content outpulls every other page type on the site except the homepage itself.
7. Cite this data
Everything on this page is published under a Creative Commons Attribution 4.0 license. Quote it, chart it, criticize it — just link back. The machine-readable version (JSON and CSV) is on the methodology and download page.
Suggested citation
PartnerCheck (2026). Dating Trust Index 2026: Global Relationship Suspicion Search Trends. Aggregated, anonymized data from 14,423 visits across 136 countries, February 24 to August 22, 2026. https://partnercheck.app/dating-trust-index-2026/
8. The limits of this data
This index measures the visitors of one website, not the population of any country. People who land here are pre-selected: they already suspect something and already chose to look for a tool. Nothing here can estimate how common infidelity is — for that, see our sourced statistics roundup, which relies on representative surveys. Geography is unavailable for about a third of (mostly early) visits, August 2026 is a partial month, and a sample of this size describes trends, not absolutes. We publish it because even a narrow, honest window into how people privately handle suspicion is more useful than the recycled, unsourced numbers that usually circulate on this topic.
FAQ
What is the Dating Trust Index?
The Dating Trust Index is an original dataset of aggregated, anonymized statistics about relationship trust searches, published by PartnerCheck. It is built from 14,423 recorded visits across 136 countries between February 24 and August 22, 2026, and covers monthly volume, country distribution, device split, hour-of-day patterns and referral sources. It contains no personal data.
When do people search for answers about a partner most?
In the 2026 index, activity peaks between 12:00 and 17:00 UTC — lunchtime and early evening in Europe, morning to midday in the Americas. The single busiest hour is 14:00 UTC (892 visits); the quietest is 05:00 UTC (260 visits), a 3.4x difference.
How much relationship-trust traffic comes from AI assistants?
About 14% of human visits in the index arrive via AI assistants — ChatGPT alone accounts for 13.8% and Perplexity for 0.1%. By comparison, the traditional search engine Bing accounts for 0.2%, meaning AI assistants refer roughly 89 times more visitors than Bing in this niche.
Is the Dating Trust Index data free to use?
Yes. The dataset is published under a Creative Commons Attribution 4.0 (CC BY 4.0) license. You may quote, chart and adapt it, including commercially, as long as you credit PartnerCheck and link to the index or methodology page. JSON and CSV downloads are available.
Does the dataset contain personal data?
No. The dataset contains aggregate counts only. Any group smaller than 50 records is suppressed, and IP addresses, session identifiers, emails, user agents, cities and scan content are never included. Uploaded photos are deleted automatically within 36 hours and never appear in any statistic.
