Methodology

SexToysDB is a desk-research database. We don’t test products, and we don’t review how they feel. Instead we join public records that are rarely put side by side: a toy’s FCC filing, the companion app’s trackers and permissions, the app’s privacy policy, and the company that actually builds the hardware. Every fact on a research card links to the source it came from.

This page explains how each part works, so you can check our work.

What a research card contains

A research card is a page about one toy. It only shows facts that have a source. Each source is recorded with:

  • the URL we read it from,
  • the date we retrieved it, and
  • an archived copy (for example on the Internet Archive) when we could make one.

When we can’t verify something, we leave the field empty and mark it unverified. We never fill a gap with a guess, an estimate, or text generated from other fields.

FCC filings: who really makes this

Every wireless device sold in the United States needs an FCC equipment authorization. The filing is public. It names the grantee, the company that applied for the authorization, which is usually the real manufacturer. Filings often include the user manual and internal photos.

How we match a toy to a filing:

  1. We look for the toy’s FCC ID in evidence: the official user manual, a photo of the product label, the brand’s own support pages, or a user manual exhibit in the filing that names the product.
  2. We record how we made the match (for example “model number on the label matches the filing”) and link the evidence.
  3. We give every match a confidence level:
Confidence When we use it
Confirmed The product name or model number appears in the FCC filing itself or on an official brand document that gives the FCC ID.
Likely The evidence points to this filing, but the product name or model number doesn’t appear in an official document. “Likely” is always labelled on the page.

If we find no filing, the card says so. Some toys are sold without a US filing, or under an ID we couldn’t trace.

We link to the official FCC record. We don’t re-host filing documents.

Same factory, different brand

When two toys from different brands list the same FCC grantee or the same manufacturer, we link them. We describe that link exactly as the evidence supports: “same FCC grantee” or “same manufacturer.” We never say two products are identical, because a shared factory doesn’t prove the products are the same.

A manufacturer link can come from:

  • the FCC grantee on the filing,
  • the brand’s own disclosure, or
  • the manufacturer’s own product listing.

Each link carries the same confirmed/likely confidence level as an FCC match, plus its evidence URL.

App privacy audits

For each companion app we record:

  • Trackers and permissions, from the Exodus Privacy report for the Android app. Exodus scans the app package for known third-party tracking libraries and lists the permissions it requests.
  • Account required? From the store listing, the app’s first-run screens or the privacy policy, with the sentence quoted.
  • Can you delete your data? In the app, by request, not at all, or unknown, with the sentence quoted.
  • Does the policy allow sharing data for advertising? Yes, no or unclear, with the sentence quoted.
  • Last store update, and whether the app looks maintained, stale or discontinued.
  • Does the toy work without the app? Only from the official manual or support pages (for example, a button that cycles modes). Otherwise we mark it unknown.

We archive the privacy policy we read, so you can see the exact version our reading was based on.

An audit describes what an app can collect and share, based on public evidence. It isn’t a security test, and it says nothing about whether the toy is any good.

The app privacy grade

The grade is arithmetic on countable, sourced facts. There are no judgment calls: two people with the same audit data get the same grade. The rubric below is version `2026-10-draft1`. If the scoring ever changes, the version changes, and each grade shows the version it was calculated with.

Points are penalty points, so lower is better.

Fact Best Worst
Third-party trackers 0 → 0 1–2 → 10 3–5 → 20 6 or more → 30
Counted “dangerous” permissions 0 → 0 1–2 → 10 3–4 → 20 5 or more → 30
Account required no → 0 optional → 5 unknown → 10 yes → 15
Data deletion in app → 0 by request → 5 unknown → 10 none → 15
Ad sharing in the privacy policy no → 0 unclear → 10 yes → 20
Updated in the last 18 months yes → 0 unknown → 5 no → 10
Total penalty Grade
0–10 A
11–25 B
26–45 C
46–70 D
71 or more F

Notes on the rubric:

  • Permissions. “Dangerous” means Android’s own dangerous protection level. We don’t count the permissions an app needs just to talk to a Bluetooth toy: BLUETOOTH_SCAN, BLUETOOTH_CONNECT, BLUETOOTH_ADVERTISE, ACCESS_FINE_LOCATION and ACCESS_COARSE_LOCATION (Android 6–11 requires location permission for Bluetooth scanning). Background location is always counted, because Bluetooth control never needs it.
  • Unknown answers count. If we can’t find whether an app requires an account or lets you delete data, that scores as “unknown.” Not being able to find it is itself a finding about how transparent the app is.
  • No report, no grade. If there’s no Exodus report for an app (for example, it’s iOS-only), we can’t count trackers or permissions, so we don’t grade it. The page says “Not graded.” We never estimate missing counts.
  • Each app page shows the full breakdown, one line per fact, so you can redo the sum yourself.

Right of reply

Before we publish a C, D or F grade, we send the brand our findings and give them 7 days to respond. If they reply, we check what they send, correct anything we got wrong, and publish their response on the app page. See [Corrections](/corrections/) for details.

Review before publishing

Research records are drafted with help from AI agents that collect the source links, FCC records and Exodus reports. Nothing is published until {{EDITOR_NAME}} has opened every source link and approved the record. Records without a source for each fact are rejected. AI is never used to write product descriptions or opinions for our pages. Page text comes from templates filled with the sourced fields.

How often we re-check

Each card shows the date it was last researched. We re-check app audits at least every six months and when an app gets a major update, and we re-check FCC and manufacturer links when a brand releases a new version of a toy. When an app stops being updated or is pulled from the stores, we mark it stale or discontinued and list it on the app shutdowns tracker.

What we don’t do

  • No hands-on testing. We don’t own or use the products. We won’t tell you how something feels, how loud it is in practice, or how long it lasts.
  • No paid placements. Brands can’t pay to be listed, to change a grade, or to be left out.
  • No invented facts. No made-up FCC IDs, trackers, grades or dates. Unverified means unverified.
  • No ratings we didn’t earn. We removed the old formula-based scores (Stealth, Durability, Value, Travel, Satisfaction), because they were calculated from copied manufacturer specs and looked like ratings we hadn’t earned.

How we make money

Some store links on SexToysDB are affiliate links. If you buy through one, we may earn a commission at no extra cost to you. Affiliate relationships never change a research card, a grade, or which products we research. See the Editorial Policy.

Found a mistake? Tell us on the Corrections page.