Substack Signup Form Conversion: The Best Convert 7x the Worst

The best Substack signup forms convert about seven times better than the worst. Same platform, same kind of publication, same widget.

That is the finding. The median form converts 0.99% of the people who see it, the bottom quartile converts 0.32%, and the top quartile converts 2.15%. Stretch to the tails and the top 5% of forms clear 10.42%, which is thirty times the bottom quartile.

Measured across 3,073 live embeds and 4.7 million views, between 2 July and 30 September 2026. As far as I know it is the only published conversion data on Substack signup forms specifically, so the method is below in full, including the part where the obvious calculation turned out to be wrong and the part where my own attempt to make the numbers look better backfired.

#The spread is the story

	Conversion of views	Share of forms
Bottom quartile	0.32%	—
Median	0.99%	49.7% clear 1%
Top quartile	2.15%	25.8% clear 2%
Top decile	5.74%	11.9% clear 5%
Top 5%	10.42%	5.7% clear 10%

159 subscribe forms with 300+ views each.

Read the right-hand column, because it is the one you can act on. Half of all signup forms fail to convert 1% of the people who see them. A quarter clear 2%. One in twenty clears 10%.

So "is 1% good" is the wrong question. 1% is the middle of the pack. The useful question is what the one-in-twenty are doing, and the answer is almost never the form. At those rates the traffic is already qualified: a reader finishing a post, clicking a link you promised something behind, arriving from a recommendation. The form converts well because the decision was made upstream.

Which is why the spread matters more than the median. A sevenfold gap between quartiles is far larger than the gap between any two form designs. Placement and traffic quality are doing nearly all the work.

#Feed widgets: one in seven clicks through

This is the cleanest number on the page and the one I would pay most attention to.

A feed widget shows your recent posts on someone else's page. It captures no email addresses at all, so on a signup dashboard it looks like it does nothing. In fact roughly one in seven people who see one clicks through to read a post. Among widgets with at least 300 views, 38% exceed a 10% click-through rate and 21% exceed 20%.

I am quoting the pooled figure here rather than a median, and unlike the subscribe numbers that is defensible: feed traffic is spread across thousands of sites. The single largest widget is 10% of views and the top fifty are 51.8%, against 71.7% for one embed on the subscribe side. There is no dominant outlier to distort it.

It also reframes what the widget is for. It is a traffic mechanism, not a capture mechanism, and judging it on signups measures the wrong thing entirely.

#Popups beat inline forms, which surprised me

	Bottom quartile	Median	Top quartile
Popup	0.79%	1.19%	2.63%
Inline subscribe form	0.32%	0.99%	2.15%

The popup wins at every point in the distribution, and its floor is more than double the inline form's.

I did not expect that, partly because popups have a bad reputation and partly because the pooled figures say the opposite. Two honest caveats: the popup sample is 15 embeds, so this is directional rather than settled, and a popup buys its rate with attention it takes from the reader, which a form sitting quietly at the end of a post does not. A 1.19% median is not automatically worth the interruption.

But the common claim that popups do not work is not supported here.

#Why the obvious number is wrong

Here is the part most benchmark posts leave out, and the reason to trust the rest.

Divide total submissions by total views across every subscribe form and you get 0.632%. Real calculation, real data, and close to useless, because embed traffic is wildly concentrated:

• One single embed accounted for 71.7% of all views
• The top 10 embeds accounted for 92.4%
• The top 50 accounted for 97.1%

A pooled average across that distribution is not a benchmark. It is one large publication's conversion rate wearing a benchmark costume. And it is wrong in a specific direction: that dominant embed converts below the median, which is exactly what drags the pooled figure under what a typical form achieves.

The fix is the median per embed, with a minimum traffic threshold so a form with nine views and one signup is not recorded as an 11% conversion rate.

#Where I tried to make the numbers look better and failed

The threshold is the one judgement call that could be used to flatter the results, so here is what happens across all of them.

Minimum views	Median	Top quartile	Forms in sample
300	0.99%	2.15%	159
500	1.01%	2.05%	115
1,000	0.73%	1.74%	77
2,000	0.60%	1.60%	51

I expected a higher threshold to raise the median by stripping out noise. It does the opposite. Higher-traffic embeds convert worse, because they sit on larger sites with colder, broader audiences. A niche site with 400 views a month has a more engaged reader than a busy one with 20,000.

So the most inclusive threshold is also the most flattering one, which is a happy accident rather than a decision. It is reported at 300 views because that is the largest sample, and the table above is here so you can see the alternative rather than take my word for it.

That finding is worth more than the benchmark, incidentally. It means traffic volume and conversion rate trade off against each other, and a falling conversion rate as your traffic grows is normal rather than a problem.

#Method, in full

• Window 2 July to 30 September 2026, 90 days.
• Population 3,073 embeds across 4,731,753 recorded events, being every live Supascribe embed serving traffic in the window.
• Conversion submissions divided by views, calculated per embed, reported as the median across embeds. Never pooled, except the feed click-through, where the concentration check above shows pooling is safe.
• Threshold embeds with fewer than 300 views are excluded from rate figures, which reduces 925 subscribe embeds to the 159 in the sample. They remain in the totals. Sensitivity across four thresholds is in the table above.
• Deduplication subscriber counts are unique by email address. 18.9% of raw submissions were duplicate addresses, mostly people testing their own form after installing it, so 28,592 raw submissions represent 23,225 unique people. Any subscriber count that has not been deduplicated is overstated by about a fifth.
• Rates use raw submissions, because a duplicate submission is still a conversion event on that form.
• Excluded 20 embeds with no recorded type, showing an implausible 8.46%, almost certainly internal testing.

What I could not measure. I wanted a breakdown by website platform, so you could see whether Webflow installs convert differently from WordPress ones. The data will not support it: only one bucket reached a usable sample, because most installs run on custom domains and the hostname tells you nothing about what built the site. Rather than publish a breakdown resting on a handful of embeds each, I left it out.

#What to actually do with this

Find your quartile, then stop reading benchmarks. Under 0.32% and something is structurally wrong, usually the form is below the fold or the traffic is wrong. Over 2.15% and you are in the top quarter, where the next gain is more traffic rather than a better form. In the middle, placement is your lever.

Do not chase the top 5%. A 10% conversion rate is a traffic fact, not a form fact. You get there by putting the form in front of people who already decided, not by rewriting the button.

Expect your rate to fall as you grow. The threshold table shows that is what normally happens. Judge the absolute number of subscribers alongside the rate.

Change one thing at a time. Moving the form and rewriting the headline in the same week teaches you nothing about either.

Judge feed widgets on clicks. By design they capture nothing. 13.26% is the bar and 20% is achievable.

#The measurement problem underneath all of this

None of the above is visible inside Substack, and that is a boundary rather than a criticism. Substack sees a subscriber the moment they land on the list. It does not see what happened before.

So if you have a form in your sidebar, one at the end of your posts, one on an about page and a popup, Substack shows you four sets of subscribers that look completely identical. You cannot tell which placement produced them, which means you cannot improve any of them. Given a sevenfold spread, that is the single most expensive blind spot in newsletter growth.

Every subscribe form, feed widget and popup that Supascribe places on your site reports its own views, submissions and conversion rate, so every comparison on this page is one you can run on your own numbers. Subscribers still sync straight to your Substack list exactly as before. What Substack's own dashboard does and does not cover is in Substack analytics.

#The short version

The best Substack signup forms convert about seven times better than the worst, and the top 5% convert thirty times the bottom quartile. The median is 0.99%, half of all forms fail to clear 1%, and one in twenty clears 10%.

Feed widgets drive 13.26% click-through and capture no emails, which is the point of them. Popups edge out inline forms on a small sample.

The spread between best and worst is larger than the difference between any two form designs. Where the form sits and who is looking at it matter more than what it looks like, and neither is something Substack can show you.