How we test Findy
This page describes the testing programme we have designed for Findy: what we will test, what we will record, and how we will report it — including the failures. It does not yet contain results, because the first controlled run has not been completed.
No results published yet
We have not completed a controlled test run against a defined test set. Until we have, this page publishes the method only. We will not publish accuracy figures, detection percentages or comparative claims until there is a repeatable test set with a stated sample size behind them. When results exist, they will appear in the Results section below, and the date of the run will be stated.
Why the methodology comes before the number
An accuracy figure means nothing on its own. "Detects 97% of hidden cameras" is unanswerable unless you also know which cameras, powered on or off, on which network, in what light, at what distance, on which iPhone, and how many were tested. The same app can be made to score anywhere between very high and very low simply by choosing the test set.
So we are publishing the method first, and we are publishing it before we have anything flattering to report. If we later publish a number, you will be able to check what produced it.
If a detection app states a headline accuracy figure without a test set, a sample size and a description of the conditions, treat the figure as marketing rather than measurement. There is no industry-standard benchmark for hidden-device detection on a phone, so there is nothing for such a number to be measured against. Ask what was tested and how many times.
The test matrix
Each row below is a category of test setup. Every category will be exercised across multiple iPhone models, and each individual test will be repeated so that we can see whether a result is stable or incidental.
1. Wi-Fi cameras
- Camera powered on and connected to the same Wi-Fi network as the iPhone
- Camera powered on but joined to a separate network the iPhone is not on
- Camera powered on, same network, with client isolation enabled on the router
- Camera powered off
- Camera powered on but recording locally with no network connection
2. Infrared cameras
- Infrared illumination active, dark room
- Infrared illumination active, brightly lit room
- Infrared illumination inactive, dark room
- Infrared illumination inactive, brightly lit room
- Each of the above viewed with both the front and the rear camera
3. Bluetooth devices
- Item tracker
- Headphones
- Portable speaker
- Smartwatch
- Unknown or generic beacon with no recognisable identity
Each is tested advertising, paired to another device, and powered off, at several distances.
4. Physical lenses
- Lens fully exposed
- Lens recessed inside a housing
- Lens behind tinted or smoked plastic
- Lens partially obstructed or dusty
5. False-positive sources
These are objects with no camera in them at all. A test passes here only if Findy does not raise a warning.
- Polished screw heads and metal trim
- Status LEDs and standby indicators
- Mirrors and glossy glass surfaces
- Chargers and power adapters
- Smoke and heat detectors with no camera fitted
- Remote controls and infrared sensors
6. iPhone models
Because infrared visibility and camera behaviour depend on the hardware, every camera-based test is repeated across iPhone generations with different camera systems, and across the iOS versions we support. See device compatibility for the current status of model-specific characterisation.
What we record for every test
One row per test, and no row is discarded because the outcome was unflattering.
| Field | What it captures |
|---|---|
| Test setup | The scenario from the matrix above, including network topology and room configuration. |
| Device under test | The object being detected, or the ordinary object being checked for false positives. |
| iPhone model | The exact handset, since camera modules differ between generations. |
| iOS version | The operating-system build, since API behaviour and permissions change between releases. |
| App version | The Findy build tested, so a result can be tied to a release in the changelog. |
| Lighting conditions | Dark, dim, ordinary indoor lighting, or bright — recorded because camera methods depend on it. |
| Distance | Distance between the iPhone and the object, and whether anything was between them. |
| Feature used | Lens reflection, infrared viewing, network scan, Bluetooth discovery or magnetometer. |
| Expected result | What the method should produce if it works as documented. |
| Actual result | What Findy actually displayed, including the label and the evidence shown. |
| Limitations observed | Anything about the run that constrains what the result can be taken to mean. |
Result categories
Every test is placed in exactly one of five categories. Three of the five are ways of being wrong, and we will report all five.
Detected correctly
Correct
The device was present and Findy surfaced it with an appropriate label and accurate supporting evidence.
Not detected
Missed
The device was present and Findy showed nothing. This is the category most worth reading, because it defines where the app should not be relied on.
Detected but misclassified
Misclassified
Findy surfaced the device but described it wrongly — for example listing a camera as an unidentified device, or attaching the wrong evidence.
False warning
False warning
Findy flagged something worth checking where there was no camera, microphone or tracker at all. Screws, LEDs and chargers all belong here.
Inconclusive
Inconclusive
The run could not be scored — interference, a setup fault, or a result that did not reproduce. Counted and reported rather than quietly dropped.
Missed detections, misclassifications and false warnings will be published alongside the successes, with the conditions that produced them. A testing page that only reports what worked is an advertisement. The point of this exercise is to tell you where Findy stops being useful, so that you keep inspecting the room yourself.
Results
No results published yet
The first controlled run against the matrix above has not been completed. Nothing on this site states a detection rate, an accuracy percentage or a comparison against another product, and nothing will until:
- the test set is fixed and written down, so the same tests can be repeated later;
- the sample size is stated for every figure we report;
- the iPhone models, iOS versions and app version behind each figure are stated;
- the failure categories are reported with the same prominence as the successes.
Until then, please read what Findy cannot detect and treat the app as an inspection aid that supports your own visual check rather than replacing it.
How testing feeds back into the app
- Run the matrix against a specific app build and record every field listed above.
- Categorise each run into one of the five result categories.
- Where a false warning is common, change the wording or the threshold rather than the colour of the label.
- Where a device class is consistently missed, document it on the limitations page rather than implying the app covers it.
- Publish the run, including its failures, and note the corresponding release in the changelog.
See what each detection method actually measures · Read the complete limitations
Last reviewed 2026-07-23 · Written and reviewed by the Findy developer.