A filter is only as good as what it really blocks. That is why we measure, by hand and with numbers, whether Die Ratel blocks the sites parents expect, and whether it leaves open the sites children need. Here is how we do it, and what we found — including the things that are not good enough yet.
The test: more than 1,600 sites, checked by hand
On 20 October 2026 we put together a test set:
- 1,211 sites that must be blocked, spread over 23 categories: gambling, alcohol, drugs, weapons, dating, games, AI companions, and more.
- 404 sites that must stay open, even if they look suspicious at first glance. Think of Gamblers Anonymous, or a medical website about drug abuse.
- Child-safe sites that nobody may block: 200 South African news, school, bank, shop and government sites, plus helplines like SADAG, Childline and Lifeline.
Every site was checked directly and really exists. We chose South African sites first: local bookmakers, liquor stores, radio stations, gun shops, vape shops and dating sites. A filter that only knows American sites does not help a family in Pretoria.
What we found: 40% was not good enough
The first measurement was sobering. The filter caught only 40% of the sites that should have been blocked. Some categories were very weak: alcohol 8%, weapons 10%, drugs 16%, dating 23%.
We also found mistakes that leaned the other way:
- Gamblers Anonymous was blocked as gambling, while it exists to help people stop gambling.
- WebMD, a medical information site, was blocked as advertising.
- Two entire French blog platforms were blocked as adult content, with a primary school’s blog among them.
- Blocking search engines also blocked Yahoo Mail.
What we fixed
- Added new source lists, including the UT1 lists from the University of Toulouse.
- Added every site we checked to our own list.
- Added a rule for new sites with obvious words in the name, like casino, betting or liquor.
- Excluded sites children use in class: Kahoot, Blooket, Prodigy, Scratch and typing programs do not fall under “games”.
- Fixed the four mistakes above.
After the fixes the filter catches 99% of the test set: 1,200 of the 1,211 sites.
The honest number
That 99% is on sites we checked ourselves and added to our list, so it is easy. The harder question is: how well does the filter catch sites we have not yet seen?
That is why we keep part of the test set aside and do not add it to our list. On those sites the lists and rules catch about 45% on their own. That shows why we keep testing and updating lists, and why a parent can block a site themselves if one slips through.
Adult content: the Ratel score
For adult content we measure separately, the way independent test labs do it. We draw 1,000 real adult sites from a list we do not use ourselves, and see how many our filter blocks:
| Filter | Adult sites blocked | Child-safe sites blocked by mistake |
|---|---|---|
| Free Family filter | 49% | 0 of 193 |
| Paid child profile | 88% | 0 of 193 |
Our goal is 95%, without a single child-safe site being blocked. We are not there yet, and we would rather say so plainly.
Choices we made deliberately
- Teams, Zoom and Google Meet fall under “instant messaging”. If you block instant messaging, you also block the school’s video calls, unless you allow that app. That is a choice for the parent, not a mistake.
- Most harm-reduction sites about drugs, which give young people information to stay safe, are not blocked as drugs.
- A DNS filter blocks whole sites, not single pages. Sometimes someone uploads spam with adult text to a large organisation’s website. We block the sites it links to, but not always the spam page itself.
Test it yourself
Open dieratel.com/check on any device. The page tries to load a test site that every filter blocks, and shows immediately whether that device is protected.
We measure again every time we update the lists. Die Ratel blocks adult content, gambling, AI companions and more on the home Wi-Fi and on the children’s phones. You set it up with our assistant, in Afrikaans or English.