Landing page conversion rate is conversions divided by visitors, times 100. The median landing page in Unbounce's benchmark of more than 41,000 pages converts 6.6% of its visitors, but that number is only useful if your own rate was counted the same way, from enough traffic, against the right comparison. Most of the arguments we see about whether a page "converts well" are really arguments about the denominator.
This guide covers the formula and its variants, what a good rate looks like by industry and traffic source, and the sample-size arithmetic that tells you whether your number is real. The benchmarks come from Unbounce, LocaliQ and WordStream, and First Page Sage; the measurement rules come from Google's own documentation.
The landing page conversion rate formula
The basic formula has two inputs and one rule: both inputs must cover the same page, the same dates and the same traffic.
Conversion rate = (conversions ÷ visitors) × 100
If a demo page received 2,000 visitors last month and 70 of them submitted the form, the conversion rate is 70 ÷ 2,000 × 100 = 3.5%. Google Ads uses the same shape with different inputs. Its help centre gives the example of 50 conversions from 1,000 interactions, which is a 5% conversion rate.
Define the conversion first. Count one action the page exists to produce: a form submission, a booked meeting, a trial sign-up. Adding newsletter sign-ups to demo requests flatters the rate.
Count completions, not attempts. Fire the event on the thank-you state, not on the submit click, which also fires when validation rejects the form.
Remove what is not a lead. Spam, internal tests and duplicate submissions inflate the numerator. Filter them, and write the filter down so next month's number is comparable.
The four denominators, and what each one tells you
"Visitors" sounds obvious until you open an analytics tool and find four candidates. Each answers a different question.
| Denominator | Count | Conversions | Rate | What it answers |
|---|---|---|---|---|
| Ad clicks (interactions) | 1,400 | 35 | 2.5% | Is the ad spend producing leads? |
| Page views | 1,600 | 35 | 2.2% | Rarely useful: reloads and back-button visits inflate it |
| Sessions | 1,250 | 35 | 2.8% | Does a single visit convert? |
| Users | 1,000 | 35 | 3.5% | Does a person convert, over however many visits? |
Source: Octo worked example, 35 clean leads after removing 5 spam and duplicate submissions from 40 raw form submissions; metric definitions from Google Analytics and Google Ads Help
Sessions. GA4's session key event rate is "the percentage of sessions that converted". A visitor who returns three times before booking counts as one conversion against three sessions. This stricter number suits paid campaign pages, where most people arrive once.
Users. GA4's user key event rate is "the percentage of users who converted". It is higher, because returning visits collapse into one person, which suits B2B pages where a buyer comes back a week later. First Page Sage's B2B benchmarks use unique visitors for this reason.
Ad clicks. Google Ads divides conversions by "ad interactions", so it covers paid traffic only, and some clicks never load the page. Expect it to differ from GA4, and do not average the two.
Page views. Avoid it: refreshes inflate it whenever the page is slow or the form errors.
What GA4 changes about the count
GA4 now calls the actions you mark as important "key events", and reserves "conversions" for the subset shared with Google Ads. Mark the form completion as a key event, then read the key event rate filtered to that landing page.
Known bots are already removed. GA4 excludes known bots automatically, using the IAB's International Spiders and Bots List plus Google's own research, and you cannot turn that off. Unknown bots are another matter: Imperva's 2025 Bad Bot Report found automated traffic made up 51% of all web traffic in 2024, and bad bots 37%, up from 32%. A jump in sessions with no matching rise in leads is often bots diluting the denominator.
Consent changes the denominator too. GA4 only models visitors who decline analytics cookies once a property logs at least 1,000 denied-consent events a day for seven days and at least 1,000 daily consenting users. Most landing pages never reach that, so GA4 counts consenting visitors only, while your form tool counts every submission. That is why form submissions divided by GA4 users can overstate the rate.
What counts as a good landing page conversion rate
A good rate is one that beats the median for your traffic source and page type, measured the same way. The industry figures below are medians, so half of all pages sit below them. Our separate post, Landing Page Conversion Rates by Industry 2026, breaks these down by sub-category; here they are as a quick reference.
| Benchmark | Rate | Basis |
|---|---|---|
| All-industry landing page median | 6.6% | Unbounce, 41,000+ pages, 57 million+ conversions |
| Industry medians, lowest to highest | 3.8% to 12.3% | Unbounce, nine industries |
| SaaS landing page median | 3.8% | Unbounce |
| Google Ads search, all industries | 8.18% | LocaliQ and WordStream, 2026 |
| Google Ads search, business services | 4.85% | LocaliQ and WordStream, 2026 |
| Google Ads search, finance and insurance | 2.64% | LocaliQ and WordStream, 2026 |
| B2B SaaS website, visitor to lead | 1.1% | First Page Sage client data, 2022 to 2025 |
| Legal services website, visitor to lead | 7.4% | First Page Sage client data, 2022 to 2025 |
Source: Unbounce Conversion Benchmark Report; LocaliQ search advertising benchmarks, updated 2026-06-01; First Page Sage B2B conversion rates by industry, updated 2025-09-18
All three are vendor-sourced. Unbounce's figures come from pages hosted on Unbounce, mostly single-goal campaign pages. LocaliQ's are click-based ad-platform rates from its own customers. First Page Sage's are whole-website rates from its clients, with no site count published.
Use the right row. A B2B SaaS demo page fed by paid search should be compared with the SaaS paid search rate, not the 6.6% all-industry median. Unbounce's SaaS page puts SaaS paid search at a median of 4.1%, email at 16.9% and paid social at 2.9%.
Traffic source moves the rate more than the page
Across all industries, Unbounce found email visitors converted at an average of 19.3%, paid social at 12% and paid search at 10.9%. Email traffic converts 77% better than paid search on those numbers. The same page, unchanged, can therefore report 4% one month and 9% the next because the campaign mix changed.
Segment before you judge. Read the rate by session source or medium before reading it as a whole. A blended rate that falls while every segment holds steady means more cold traffic, not a worse page.
Device matters less than people assume. Unbounce reports that 83% of the visits it analysed came from mobile, and desktop converted 8% better than mobile on average. That gap is real but small next to the gap between channels.
Copy matters more than people assume. Pages written at a fifth to seventh grade reading level converted at 11.1% in the Unbounce data, against 5.3% for pages written at a professional level. Word count and reading time both correlated negatively with conversion. That is a correlation across many pages, not a controlled test, but it points the same way as most landing page audits.
How many visitors before the rate means anything
A conversion rate is an estimate. With few visitors, the estimate is wide. The Wilson score interval, which statisticians prefer to the simple normal approximation when rates are low or samples small, gives the range below at 95% confidence.
| Visitors | Conversions | Measured rate | Plausible true rate (95%) |
|---|---|---|---|
| 100 | 4 | 4.0% | 1.6% to 9.8% |
| 300 | 12 | 4.0% | 2.3% to 6.9% |
| 1,000 | 40 | 4.0% | 3.0% to 5.4% |
| 10,000 | 400 | 4.0% | 3.6% to 4.4% |
Source: Octo calculation, Wilson score interval at 95% confidence, formula as given on Wikipedia's binomial proportion confidence interval article
A page that reports 4% on 100 visitors has not shown it is below the 6.6% median. A page at 4% on 10,000 visitors has. To narrow the estimate to within one percentage point either side, you need roughly 1,475 visitors at a 4% rate, and about 5,900 to get within half a point (normal approximation, 95% confidence).
Comparing two versions needs more. Detecting a change between two rates takes far more traffic than measuring one rate. Evan Miller's sample size calculator, at its default settings of a 10.2% baseline and a three-point minimum detectable effect, asks for 2,545 visitors per variation. At a 3% baseline the number climbs sharply. Optimizely's analysis of 173,000 experiments found only 12% produced a statistically significant win on the primary metric, which is a useful reminder that most changes do not move the rate at all.
How to read your own rate, step by step
This is the order we would follow when a client asks whether a landing page is working.
- Fix the definition. One conversion type, counted on completion, spam removed.
- Fix the denominator. Sessions for one-visit campaign pages, users for long B2B consideration. Keep it the same every month.
- Check the traffic. Session spikes with no matching conversions are possible bots.
- Segment by source. Compare each channel with that channel's benchmark.
- Put a range on it. Under about 1,000 visitors, report the plausible range, not one number.
- Compare with yourself first. Last quarter's rate, same definition, beats any industry median.
- Only then change the page.
Calculator shortcut. With conversions in column A and visitors in column B, A2 divided by B2 gives the rate. For a rough 95% range, add and subtract 1.96 times the square root of rate × (1 minus rate) ÷ visitors; this normal approximation misleads below roughly 1,000 visitors, where the Wilson interval above is safer.
What this means for judging a landing page
Before deciding a page underperforms, check that the conversion is counted once, the denominator matches the benchmark, the traffic is human, and there is enough of it. Pages that are really broken show it consistently, per channel, over a few thousand visits. When we build a marketing site in Webflow, we would set the key event, the spam filter and the per-channel report up before launch, so the first month's number can be trusted. Our marketing websites service page explains how we scope landing page work.
