Sitecomb research

How our research works

We publish what we find when we check many small-business websites at once.

This page says how we pick them, what we check, and what we will and won't publish.

Where the websites come from

We start from OpenStreetMap (© OpenStreetMap contributors, ODbL): the free, public map that volunteers keep up to date.

We take independent businesses that list a website.
We group them by town and by kind of business: food and drink, health, beauty, trades, shops, professional services, education, community.
We pick a random sample from each group.
We don't scrape Google, Yelp or any other directory.

OpenStreetMap lists the businesses that mappers have added, not every business. So our numbers describe the businesses in our sample, and we say so.

Each website gets the same check as a Sitecomb free score, run one site at a time so no website is put under load.

Our checker reads public pages only, respects robots.txt and says who it is. What our checker does.

Some sites can't be checked: they block automated checks, they're down, the address no longer exists, or the page doesn't load.

We count those separately and never as passing or failing.

We look up the website's DMARC record, a public setting that tells inboxes what to do with email that pretends to come from that domain.

Passed means it tells inboxes to quarantine or reject fakes.
A record set to monitoring only, or no record at all, didn't pass.
We only count sites on their own web address.
On a shared address (such as name.example-builder.com), the setting belongs to the platform, not the business.

We publish only checks that are right at least 95% of the time when we test them against known answers.
Every percentage is cut down to one decimal, never rounded up.

Only totals and shares for groups of websites. We never name a business, publish a website's address, or its score.
A town or kind of business is broken down only when the group has at least 20 websites.
For groups under 50, we don't publish a share under 10% or over 90%, so a small group can't point at the businesses in it.
We say “didn't pass our check on [date]”, never “broken” or “insecure”.
Every article gives the dates, the sample size, the source and a link to this page.

The result for each website is kept privately and used only for this research.

If you own a site in a sample and want it left out of the next update, email hello@sitecomb.com and we'll take it out.

We checked Ireland (October 2026): 919 websites, 14 towns, checked 5–6 Oct 2026.