Landing Page Conversion Rate: How to Tell Whether Your Number Is Decision-Ready

2026-09-15

Landing page conversion rate is conversions divided by the people who arrived, expressed as a percentage. If 400 people landed and 8 of them did the thing you wanted, that is 8 ÷ 400 = 0.02, or 2%.

That is the whole calculation, and it is the easiest part of the job. The harder question is whether the number you just produced is good enough to support a decision. Published averages for this metric differ by several multiples depending on who measured, when they measured, and what they counted as a conversion. So a benchmark can tell you your number is lower than someone else's. It cannot tell you why, or what to do next, or whether your number is measuring what you think it is.

What follows are the checks that answer that. They run in the order a funnel actually runs: whether the data is real, whether the right person is arriving, and whether the path from arrival to action holds together. Each of those stages is a precursor to the ones after it. If a stage fails, fix it before you rely on anything you concluded at a later one.

First, define the event and the denominator

Write down exactly what behaviour counts as a conversion and exactly what population it is divided by, before you use the rate in any decision. A page that counts "clicked the button" and a page that counts "submitted the form and the record reached the CRM" produce very different percentages from identical performance.

The denominator matters just as much. Unique visitors, sessions, and page views are three different numbers, and dividing by the largest one makes any page look worse. If a rate is being quoted and nobody can say what it counts or what it divides by, you do not have a metric yet — you have a figure.

Skip this if every number in play already has a written definition with an explicit denominator. Otherwise it is the first thing to fix, because everything below depends on it.

What to do with the benchmark you found

When you encounter a specific figure — an industry average, a "good" rate, a threshold — record where it came from and what produced it, then verify it against your own current data or do not use it. A number of unknown origin shaping a real decision is a risk you are taking without knowing you have taken it. Your own tracked performance, with a definition attached, outranks any external average. If you are working from your own current, verified data, this one is already handled.

And do not average two populations into one rate. Cold paid traffic and warm email traffic on the same URL, two offer versions in the same reporting window, or a period spanning a campaign objective change — blend those into one number and that number cannot tell you which one moved. Keep the combined total for operational reporting if you like. Do not use it as a test result. If there genuinely was one source, one page version, and one offer in the period, this does not apply to you.

Second, establish that the number is real

Before interpreting any performance data, run a check on the mechanics: date range, timezone, filters, campaign status, destination URL, which variant is actually live, event configuration, and sample size. This is unglamorous, and it is the step that stops you optimising a page while the traffic has been going to an old URL. You can skip it if you already ran it on this data set and nothing has changed since.

Fire the event yourself

Complete the conversion action on a real device and watch for the event to arrive. If it never lands, the rate you have been optimising against was never measuring that action. Do this before concluding anything from a low number or a zero — especially if a variant is live and reporting nothing, or conversions dropped off a cliff. Verification like this needs one successful test, not a sample. If the event has been verified since the last change to the page, the funnel, or the tracking, you already have your answer.

Draw the actual technology path a visitor takes and assign a verified event to each point where the domain, platform, or tool changes. Check, too, whether your thank-you page is standing in for something it does not actually prove.

Reconcile the count chain

Line up impressions, clicks, page views, sessions, form starts, and completed actions for the same period, and investigate the largest unexplained gap. This is what separates "my page does not convert" from "people are not reaching my page." If the chain has already been reconciled for this period and the gaps are explained, move on.

Where clicks are not becoming page views, measure load time on a real mobile device and check redirects and script placement before you attribute weak engagement to the copy. A behaviour tool showing far fewer sessions than the ad platform shows clicks is a gap worth explaining before you rewrite a paragraph. If your click-to-page-view ratio is already healthy and mobile load times are known to be fine, this is not your gap.

Decide which system is authoritative

For every outcome, name the system that actually receives it — the CRM, the calendar, the checkout, the payment processor — and reconcile platform-reported conversions against it. When the ad platform and your own records disagree, the disagreement itself is the finding. If the two already agree within a margin you understand, this one is settled. The same discipline applies further down the funnel; we covered it in more detail in how to tell whether your pipeline numbers are real.

Third, check you have enough events to conclude anything

Count the events, not the percentages, and ask what one more event would do to the number. Eight conversions in 400 sessions is 2.0%. Nine in 400 is 2.25%. That quarter of a point shows you exactly how much weight a single form fill is carrying in the figure you are about to act on — which is the thing a percentage hides and a raw count does not.

Then decide how much evidence this particular decision needs before you look at the result, based on how expensive and how reversible it actually is for you. Relaunching a paused variant and commissioning a rebuild are not the same decision, and neither is universally cheap or universally costly — price your own case rather than assuming. Confirming that an event fires sits at one end of that range; something expensive or hard to undo sits at the other.

One more timing trap: if buying takes weeks, link each sale back to the enquiry that started it and report cohorts that have not finished converting separately from matured ones. If the cohort has not matured, you are reading a half-finished result and calling it a verdict. Where purchase is effectively immediate, same-period comparison is valid and this does not apply.

Fourth, check whether the page is where the loss is

The page sits late in the chain. Whether the right person understands the offer, and whether the right person is arriving at all, both come before it — and a problem at either point can depress the rate on your dashboard while the page itself is doing its job.

So when engagement is strong and downstream conversion is weak — healthy click-through, poor opt-ins — investigate who is arriving before rewriting anything. Did the audience or geography change before the drop? Is cheap traffic being bought because it is cheap? Ask those first. If engagement and downstream conversion moved together, the ad and the traffic are telling the same story and you can stop treating them as separate suspects. There is a fuller walk-through of that split in our guide to diagnosing ads that are not converting.

Two page-side checks belong in the same pass, each with its own boundary:

• Promise continuity. Where healthy clicks sit alongside weak page conversion, check whether the page carries the ad's core language and promise above the fold, so the visitor recognises it as the continuation of what they clicked. A new creative angle pointed at an unchanged page is one of the setups worth checking here. If the ad and page already carry the same promise to the same person, this is not your gap.
• One source, one job. Define the traffic source, what the visitor just saw, and the single action this page must sell — then remove everything that does not serve it. Paid traffic landing on a general company page, or one page serving several sources, produces a rate that averages unrelated jobs together. If the page is already purpose-built for this source and this action, skip it.

When several stages all look improvable, find the first point where a meaningful share of intended visitors leave and work there — not the section that irritates you most, and not the one nearest the money. If you have already located and addressed the earliest loss, that work is done.

Fifth, working the page itself

Page work earns its keep when your own evidence points at a consequential loss on the page. One thing here does come first: repair outright breakage before you interpret any of the data. The rest are checks to run where their condition applies, not a fixed running order.

• Repair obvious defects first. Walk the page and the whole path on real devices, looking for things that are simply broken — layout faults, styling failures, a button that does nothing. Fix those before drawing any conclusion from the numbers. If the path has been walked on real devices since the last change and nothing is broken, move on.
• Design mobile as its own experience. Open it in a real mobile browser and set the mobile order, image sizes and spacing deliberately instead of inheriting them from desktop. Desktop columns that compress into unreadable blocks, or a hero image that pushes the headline below the fold, are decisions you inherited rather than made. If mobile has been inspected on a real device and designed for, skip it.
• Test the first screen for comprehension. Can a visitor tell why they are here, what this is, and what to do next, within a few seconds and without decoding anything? Ask someone with no context to summarise the page after a glance. If they can state its purpose and next step immediately, that is not where your loss is.
• Establish the message was consumed before judging it. If persuasion rests on a video or a long section, find out how many people started it and how far they got before you rewrite it. Play rate and retention are two different numbers. Where consumption data already shows the message was received, the content itself is a fair target.

Scroll maps, attention maps and recordings belong here as a way of generating a hypothesis about where something breaks — then test that hypothesis against the actual outcome before you change anything. Check the click map before you conclude anything from a scroll map, and reconcile both against conversion. They tell you where to look, not what to conclude. And where conversion is strong, an unflattering behaviour picture is not a problem to solve.

A rate is not a result

The last thing worth saying about landing page conversion rate is that it is a middle metric. A page converting at 2% with customers worth several hundred each can be a better business than a page converting at 8% on something that barely covers fulfilment.

Model the whole customer — price, margin, fulfilment cost, any add-on or later offer, retention, refunds — and set your acquisition targets from that model rather than from an external benchmark. That is what lets you read the page's output in economic terms: whether the conversions it produces are arriving at a cost your margin supports. If you already have a current model and real cohort data tracking against it, you have the number you need. This is the same discipline that makes an acquisition cost figure decision-ready rather than merely available.

If these checks do not isolate it

Working through this sequence does not prove your page is fine by elimination. Thin data, execution problems, timing, seasonality, and interactions between several of these factors can all survive every check listed here. If you reach the end without isolating a cause, the honest read is that the evidence you currently hold is not sufficient to diagnose it, and the next step is more first-party investigation — not another round of headline rewrites against a number you cannot yet trust.

If you would rather have a second pair of eyes on where the leak actually sits, GrowthHackr's Campaign Leak Audit is free and works through this same sequence against your own funnel.