22 companies link to their community from the homepage. On Dropbox it is link 219 out of 230.

Our community link used to live in the main navigation. A redesign turned the navigation into four items, the link went into the footer, and signups fell. Nobody joined the two facts for about three weeks, because a footer link still works. It just stops being found.

That made me curious about where everyone else keeps theirs, so on 25 August 2026 I fetched forty software homepages and counted.

What I counted

For each homepage I listed every href in the order it appears in the HTML, then found the first one pointing at something with community or forum in it. Two numbers come out of that: the ordinal of the link, and its position as a percentage of all links on the page.

The obvious objection is that source order is not screen position. CSS can put the last link in the HTML at the top of the page. That is true, and it is the honest limit of this method.

In practice the correlation holds well enough to be useful. Navigation is written early, footers are written last, and a link at 95% of the way down a document is not in the header.

What came back

Twenty-two of the forty homepages link to a community or a forum. Eighteen do not link to one at all.

Where the link sits	Sites
First third of the page	7
Middle third	9
Last third	6

The median link sits at the 49th percentile. Halfway down. For most of these sites, that is the top of the footer.

The two ends of the range are worth naming.

Highest	Link	Of	Position
Atlassian	93	886	10.5%
Buffer	32	173	18.5%
Klaviyo	83	383	21.7%

Lowest	Link	Of	Position
Dropbox	219	230	95.2%
BigCommerce	212	230	92.2%
ActiveCampaign	149	174	85.6%
Notion	160	193	82.9%

Atlassian’s homepage carries 886 links, more than any other site here, and the community still appears in the first ninety. That is a decision somebody made and defended. Dropbox’s forum is eleven links from the bottom of the document.

Three links that do not reach a community

This was the part I did not expect, and it is the part I would check first on your own site.

Hotjar’s community link goes to community.contentsquare.com/private/login. Contentsquare acquired Hotjar. The community moved to the parent brand and now sits behind a login. Somebody clicking that link from the Hotjar homepage arrives at a sign-in form for a company they were not reading about.

Notion’s /community redirects to app.notion.com/community, which returned about 20 KB. That is an application shell, not a page with discussions on it.

Squarespace’s first community link is /solutions/nonprofits/community-groups, a marketing page about community organisations. Their actual forum lives at forum.squarespace.com, which is linked from elsewhere. The word collided with itself and the marketing page won the source order.

Of the twenty-two links, twenty returned 200 when fetched. HubSpot’s and Squarespace’s forums returned 403 to a plain request, which is bot protection rather than absence. Both load fine in a browser.

Run this on your own site

The whole check is one command. Replace the domain and the pattern:

curl -sL -A "Mozilla/5.0" https://yoursite.com/ \
  | grep -o 'href="[^"]*"' \
  | grep -n 'community' | head -3

The number grep -n prints is the ordinal. Divide it by the total, which you get by dropping the second grep and piping to wc -l.

Then do the two things the number cannot tell you.

Click your own link while logged out. Every one of the three broken cases above would have been caught in ten seconds by somebody who was not already signed in. Community teams are permanently logged in, which is exactly why nobody notices.

Search your homepage source for your community word. If a marketing page, a careers page or an image filename contains it, that thing may be sitting above your real link, and both a reader scanning the footer and a crawler following source order will meet the wrong one first.

The false positive I nearly published

My first pass counted twenty-three sites, not twenty-two. The extra one was Mailchimp.

Mailchimp’s homepage has no community link. What it has is an image called hero_community_based_temp_1x.webp, and a pattern looking for community inside an href found it happily.

One in twenty-three is a 4% error rate on a measurement that would have read as authoritative in a table. The fix took a minute. Noticing was the hard part, and I only noticed because the number moved between two runs of a script I thought was doing the same thing twice.

What I still cannot tell

Eighteen of the forty homepages have no community link. Some of those companies definitely run communities, and a homepage is not the only place a link can live.

What the data cannot separate is the two reasons for absence. One is a deliberate decision that the community is for existing customers and does not belong on an acquisition page. The other is that a link was there, a redesign happened, and it never came back.

From the outside those look identical. From the inside, one of them is worth a meeting.

Questions people ask about this