Where Website Builders Come From, and Who Launches

Search visitors were more than twice as likely to publish as people sent by AI assistants. Across 1,555 We.Inc signups, 10.3% of those from search engines published a site, compared with 7.3% from social media and 4.5% from AI assistants. Social signups started projects most often, at 79.2%.

Signups by source, Mar 1 to Oct 27, 2026
Source	Signups	Started a project	Published
Direct, other or unknown	898	443 (49.3%)	126 (14.0%)
Search engines	437	263 (60.2%)	45 (10.3%)
Social media	96	76 (79.2%)	7 (7.3%)
AI assistants	66	28 (42.4%)	3 (4.5%)
We.Inc site links and email	58	28 (48.3%)	4 (6.9%)

Updated October 28, 2026

Visitors referred by AI assistants are a fast-growing source for many sites, ours included. We compared how far they get once they sign up with people who arrive from search, social posts and ads, or a typed address.

Publish rate by source

Share of signups that published a site, by source (n = 1,555)
	Value
Direct, other or unknown	14.0% (126 of 898)
Search engines	10.3% (45 of 437)
Social media	7.3% (7 of 96)
We.Inc site links and email	6.9% (4 of 58)
AI assistants	4.5% (3 of 66)

Method and date range

Numbers come from a single read-only SQL query on our production database, executed October 28, 2026, over accounts created between March 1 and October 27, 2026.

n = 1,555 signups after exclusions. Source rules run in this order: AI assistants, search engines, social media, direct, our own site and email, then other. Started means the account owns at least one project; published means one of its projects has a publish time.

Employee accounts, internal flags, test-like and plus-tagged addresses, and throwaway bot mail domains were excluded before counting. The smallest source (8 signups) was merged into the unknown row to keep every group at 20 or more.

Limitations

Source tracking started in May 2026, so earlier signups all fall in the unknown row, inflating its rate.
Publish counts for AI assistant, social and email groups are below ten, so their rates are rough.
Offers and pricing changed during the window and affected groups that arrived at different times differently.

Cite this study

We.Inc, "Where Website Builders Come From, and Who Launches", October 28, 2026, https://we.inc/research/where-website-builders-come-from.

Starting is not finishing

Social visitors had the highest start rate and one of the lowest publish rates. Most came from short ad campaigns in August to early October, where the prompt box was the first thing they saw.

Search visitors sit in the middle on starting but convert best of the tracked groups at launch, which fits someone who searched for a website builder with a site already in mind.

Why the unknown group is large

Campaign tags were not stored for early accounts, and many visitors arrive with no referrer at all. Results for that row mix true direct visits with untracked ones and should not be read as a single channel.

How it works

Took customer accounts created March 1 to October 27, 2026.
Assigned each to one source using its stored campaign tag first and its referring site second, with a fixed rule order.
Checked whether the account created a project and whether any project was published.
Merged the eight signups from minor referrers into the direct, other or unknown row so every group has at least 20 accounts.

Tips

If you run a builder or any sign-up product, measure sources to the launch step, not only to signup. The ranking changes between the two.
Social signups were eager starters but rarely finished, so follow-up help after the first draft matters most for that group.
Visitors from AI assistants often arrive mid-research, comparing tools. They started projects less often than any other group.

Frequently asked questions

Do visitors from AI assistants convert well?

In our data they signed up but rarely launched: 3 of 66 published (4.5%) and 42.4% started a project. The group is small and young, since the first such signup was on July 29, 2026.

Which source produced the most launches?

By rate, the direct, other or unknown group at 14.0%, partly because it includes early users from before source tracking began. Among tracked sources, search engines led at 10.3%.

Which search engines and assistants are included?

Search covers Google, Bing, Yahoo, Brave, DuckDuckGo and similar engines. The assistant group covers referrals tagged or referred from the major AI chat and answer products.

Start for free or see all free tools. We.Inc is free to start, with no card needed. AI building is on paid plans from $20 a month.