Blog · 24 October 2026

What is a good conversion rate? Benchmarks, with sources

A good conversion rate is a range, and the range depends on what you count. Across more than 41,000 landing pages that Unbounce studied between July 2023 and July 2024, the median page turned 6.6% of its visitors into whatever the page asked for: a sign-up, an enquiry or a sale.[1] For software (SaaS) pages the median was 3.8%.[2] On those SaaS pages, visitors from email did what the page asked at a median of 16.9%, and visitors from display ads at 0.3%.[2] So compare your number with the figure that counts the same thing, from the same kind of traffic.

The rest of this page is where those figures come from, how big the samples were, and where the evidence is old or thin.

What the published benchmarks say

“Good” and “great” are ChartMogul’s own labels for its middle and top ranges. Heap’s figure is from 2017: the sign-up forms it measured are old, so treat its level as approximate.

Three things to notice before you compare yourself with any of these.

Every row counts a different group. A landing-page rate counts visitors to one page. A sign-up-page rate counts only the people who already reached the form. A trial-to-paid rate counts only people who already started a trial. Comparing your homepage with someone else’s trial figure tells you nothing.

Asking for a card changes who starts a trial. In ChartMogul’s survey, the “good” range for trials that require a card is 25–35%, against 4–6% for free trials as a whole.[4] That rate only counts people who started a trial, and asking for a card changes who starts one. So a card-required figure and an all-trials figure describe different businesses, not a better and a worse version of one.

A median is not a target. ChartMogul notes that very few products sit at its 8% median.[4] A median tells you what the middle page or product does, not what yours should do.

Why the source of your visitors sets the range

The biggest reason two sites with similar pages get different results is where the visitors came from. These are Unbounce’s medians for SaaS landing pages, by where the visit started:[2]

Email visitors already know you; display-ad visitors mostly did not ask to see you. The gap between the two is more than 50 times. So a page at 2% on mostly display traffic may be doing its job, and the same page at 2% on mostly email traffic has a leak. Never compare one mixed number. Split by source first, then compare inside each source.

Split by device too: 79% of visits to SaaS landing pages in Unbounce’s data were on mobile devices.[2] A number that mixes phones and computers mostly describes phones.

Is a 2% conversion rate good?

It depends which 2% it is.

• As a SaaS landing-page rate, 2% is below the 3.8% median.[2]
• As a landing-page rate in any industry, it is well below the 6.6% median.[1]
• On a SaaS page fed mostly by email, it is far below the 16.9% median for email visitors.[2]
• On a SaaS page fed mostly by display ads, it is above the 0.3% median.[2]
• As a whole-site rate (every visitor, every page), we found no published benchmark with a stated sample and dates, so we give no figure. Compare it with your own past weeks instead.

Same number, different answers. That is why “is X% good?” has no single answer.

Is a 20% conversion rate good?

Often it is, but a high number deserves more checking than a low one. First ask what it is 20% of.

• 20% of a landing page’s visitors is about three times the 6.6% all-industry median.[1] Check that your own visits and repeated sends are not inflating it.
• 20% of the people who reach a sign-up form is below the 36.2% average Heap measured.[3] A page whose only job is the form should do better.
• 20% of card-required trials going paid is below ChartMogul’s 25–35% “good” range.[4]
• 20% of free trials going paid, across trials as a whole, is above ChartMogul’s 10–15% “great” range.[4]

How to find your own good number

Benchmarks give you a range. Your own numbers give you a target.

1. Pick one goal. Visit to sign-up, sign-up to first use, or trial to paid. One at a time.
2. Split by source and device. Compare each part with its own benchmark, not one mixed figure.
3. Find the step that loses the most people. That is where to look first.
4. Change one thing, then read the same numbers again. Your own trend counts for more than any industry table.

This is the job Convertlyft does on your own site. One line of code counts every visit, with no sampling and your own visits left out. Funnels of up to 6 steps count people in order, by device, against the period before, and the step that lost the most people is named. Then What to do names one change to make, with the numbers behind it. You, your developer or your AI tool makes the change; Convertlyft never edits your site’s code. More on how funnel analysis works in Convertlyft, or see the other post here: landing page optimization, and what the evidence supports.

Questions

Sources

1. Unbounce, 2024 benchmark report: 6.6% median across all industries; more than 41,000 landing pages and 464 million unique visitors. Data window from its methodology page: 23 July 2023 to 23 July 2024. Pages checked 23 October 2026.
2. Unbounce, SaaS benchmarks, same report: 3.8% median; medians by source of 16.9% (email), 5.1% (Google search ads), 2.9% (paid social), 0.3% (display); 79% of SaaS landing-page visits on mobile. Page checked 23 October 2026.
3. Heap, blog post on sign-up flow benchmarks, published 20 March 2017: 36.2% average from sign-up page to sent form, 79 SaaS companies, 27 November 2016 to 26 February 2017. Page checked 23 October 2026.
4. ChartMogul, 2026 report on free-to-paid rates, published 4 February 2026: 200 B2B software products surveyed in January 2026; 8% median free to paid; free trials as a whole 4–6% good, 10–15% great; card-required trials 25–35% good, 50–60% great. Page checked 24 October 2026.