December 2, 2025
A landing page can attract plenty of visits and still be commercially weak. Unbounce's 2024 benchmark covered more than 57 million conversions, 41,000 landing pages, and 464 million page views, finding a median conversion rate of 6.6% across industries. Yet that broad figure cannot tell you which page on your site brings the most valuable visitors.
The distinction matters even more when traffic becomes harder to earn. Contentsquare's 2026 benchmark, based on 99 billion sessions across more than 6,500 websites, reports that engagement fell 10% and conversion rate fell 5.1% year on year in its aggregated dataset. The scale of that study is impressive, but the practical lesson is local: you need to know where valuable journeys begin before you spend more money sending people there.
Luckily, this is not as complicated as it sounds. GA4 already gives you most of the raw material. The job is to combine revenue, key-event rate, session depth, and acquisition source without mistaking traffic volume for value or correlation for causation.
Think of your entry pages as doors into a shop. The busiest door is not necessarily the one used by the customers who spend the most, and a quiet side entrance may lead directly to your best department.
GA4 calls an entry page a landing page. More precisely, the Landing page dimension is the page path and query string associated with the first page view in a session. Google's Landing page report documentation lists Sessions, Active users, Key events and Total revenue among the standard metrics available for each page.
Start with a meaningful date range, normally at least one complete business cycle. For a retailer, that might be four or eight weeks. For a professional service with a longer decision process, a quarter may be more useful, provided tracking and site structure remained stable throughout the period.
Then calculate revenue per visitor using one consistent denominator:
Revenue per visitor = Total revenue / Active users
If you prefer to evaluate visit efficiency, calculate revenue per session instead. Both ratios can be useful, but they answer different questions. Active users estimate the value generated per person represented in the report, while sessions measure the value generated per visit. Do not compare one page's revenue per user with another page's revenue per session.
Total revenue in GA4 includes purchase revenue, subscriptions, in-app purchases and advertising revenue, less refunds where relevant. Google also notes in its Traffic acquisition guidance that ecommerce purchase events need both value and currency parameters for revenue to populate correctly. Before ranking anything, check that your revenue implementation is complete and uses a consistent currency.
A simple working table should include both scale and efficiency:
| Entry page | Active users | Sessions | Total revenue | Revenue per visitor | Session key-event rate |
|---|---|---|---|---|---|
| Product category A | 8,000 | 9,200 | £48,000 | £6.00 | 4.8% |
| Buying guide B | 2,000 | 2,500 | £24,000 | £12.00 | 7.2% |
| Homepage | 15,000 | 18,500 | £54,000 | £3.60 | 2.9% |
These are illustrative numbers, not a benchmark. They show why you need two rankings. The homepage wins on total revenue, but the buying guide produces twice as much revenue per visitor as the product category page and more than three times as much as the homepage.
Do not crown a page after ten visits and one large order. Set a minimum sample threshold before ranking, then show the visitor count beside every ratio. For important budget decisions, ask an analyst to add a confidence interval or a Bayesian comparison rather than relying on a league table built from tiny samples.
Revenue on a landing-page row also needs careful language. It is revenue associated with sessions that began on that page, not proof that the entry page caused every purchase. A visitor may read several pages, return later or receive help from another campaign before buying.
That being said, the ranking is still useful. High-volume, high-efficiency pages deserve protection and further investment. High-volume, low-efficiency pages deserve diagnosis. Low-volume, high-efficiency pages may be candidates for additional relevant traffic, provided the audience and message remain aligned.
Revenue is the till receipt; conversion rate shows how often visitors reach the till at all. In GA4's current language, a conversion is represented by a key event, meaning an action you have marked as particularly important to the business.
Use a specific business outcome rather than the total of every key event. A retailer might choose purchase; a clinic might choose a valid appointment request; a software company might choose a completed trial registration. Google's key-event documentation makes clear that any collected event can be marked as a key event, so the label alone does not guarantee commercial significance.
For entry-page analysis, session key-event rate is normally the cleanest starting point:
Session key-event rate = Sessions containing the selected key event / Total sessions
This is a session-level question, which fits a session-scoped landing page. User key-event rate has a different denominator and answers a different question about people during the selected period. Keep the metric definition fixed across every page you compare.
Build the comparison in the Landing page report or a Free Form exploration. Add Landing page, Sessions, the selected Key events metric, and Session key event rate. If URL parameters create multiple rows for the same content, decide whether they represent meaningful page variants or tracking noise before combining them.
A useful diagnosis separates four patterns:
| Pattern | Likely interpretation | First check |
|---|---|---|
| High traffic, high key-event rate | Proven entry route at current traffic mix | Protect message match and tracking |
| High traffic, low key-event rate | Large opportunity or poor audience fit | Segment by source, device and campaign |
| Low traffic, high key-event rate | Promising but not yet proven at scale | Increase relevant exposure gradually |
| Low traffic, low key-event rate | Weak evidence and weak outcome | Check intent, usability and sample size |
The word relevant is doing plenty of work here. A page designed to answer a support question may have no reason to produce purchases. A privacy page should not be judged against a product page, and a top-of-funnel guide may assist later outcomes without recording an immediate lead.
Compare like with like. Group pages by purpose, such as product, service, category, guide or campaign landing page, then compare equivalent outcomes. If a quote page and an educational article have different jobs, forcing both into one conversion ranking will reward the page closest to the final action rather than the page serving its intended visitor best.
Watch the trend as well as the period total. A conversion-rate drop can come from a page change, a new traffic source, device problems, broken event tracking or a shift in demand. Annotate campaign launches and site releases so you can connect movement with a real change instead of inventing a story afterwards.
Finally, inspect the experience itself. Read the page on mobile, complete the form, check error messages and compare the promise in the advert or search result with the first screen visitors see. Overoptimising a button colour while the page answers the wrong question is the digital equivalent of polishing a door that opens into the wrong room. It helps to understand what makes people click in the first place before you start rearranging the page they land on.
Session depth is a trail of footprints after the front door. It tells you where people go and how much they interact, but more footprints do not automatically mean a better journey.
GA4 defines Views per session as the number of web pages or app screens viewed per session, including repeated views. You can confirm the definition in Google's current dimensions and metrics catalogue. Pair it with Average engagement time per session, Engaged sessions and the selected key event rather than treating page count as a standalone score.
An engaged session lasts longer than ten seconds, records at least one key event, or contains at least two page or screen views. That definition means a successful one-page visit can be engaged if it produces the right action. Conversely, six page views can reflect confusion, repeated backtracking or a visitor hunting for information that should have been obvious.
Use the following scorecard to keep depth in context:
| Behaviour after entry | Possible meaning | What to inspect next |
|---|---|---|
| One page, key event completed | Fast, successful journey | Form quality and lead validity |
| Several pages, key event completed | Considered journey with useful exploration | Common supporting pages |
| Several pages, no key event | Research, uncertainty or friction | Repeated nodes, exits and missing answers |
| One page, little engagement | Poor match, fast answer or tracking gap | Source, page purpose and event setup |
Path exploration adds the sequence. Google's Landing page guidance explains that you can start a Path exploration from a landing page and inspect the pages viewed afterwards. For a high-value page, look for sensible next steps such as product detail, pricing, case studies, booking, basket or checkout.
Do not read a path diagram as a transcript of somebody's thoughts. It shows recorded sequences, not motivation. Use it to find questions worth investigating, then check page content, user feedback, support logs, session recordings where consent permits, and usability tests with real people.
Reverse the analysis for important outcomes. Start from a purchase or lead event and work backwards to see which pages commonly appear before it. This will not produce causal attribution, but it can reveal supporting content that a landing-page-only table undervalues.
Depth also needs a clean denominator. If you calculate your own metric, define whether it means total views divided by sessions, distinct pages per session, or steps completed in a funnel. Those measures are not interchangeable, and repeated views can inflate the first while a strict funnel ignores useful side journeys.
Pay special attention to (not set) landing pages and implausible routes. Sessions can begin without a conventional page view, consent settings may limit data, and tagging errors can detach events from the expected page. Fix measurement quality before you redesign a page based on a journey that never happened as reported.
The human-first test is simple: did the path help the visitor accomplish the task with appropriate confidence? In healthcare, finance, legal services and other trust-sensitive fields, extra steps may be necessary for explanation, consent or risk disclosure. Shorter is not always better, but every step should earn its place.
An entry page is like a stage, and the traffic source changes the audience in the seats. The same performance can delight people who asked for it and disappoint people who arrived after a very different promise.
Use session-scoped acquisition dimensions for this comparison. GA4 organises traffic-source data at user, session and event scope. Google's scope guide explains that First user source describes initial acquisition, while Session source describes the source associated with a particular session.
Pair Landing page with Session source / medium or Session default channel group. Do not substitute First user source merely because it is available. That would compare the page with how the person was first acquired, which may have happened weeks before the session being analysed.
For each page and source combination, show Sessions, Active users, Total revenue, revenue per visitor, Session key-event rate, Views per session and Average engagement time per session. Add Device category when a channel has a distinctive mobile or desktop mix.
Expect the results to differ. Branded search may bring people who already know you. A paid social campaign may introduce cold prospects. Email can bring existing customers, while a referral partner may send a small but highly qualified audience. The question is not which channel wins universally, but whether each source sends the right people to the right first page.
Start with a pivot table and read it in both directions. First, compare sources within one entry page to find message or audience mismatches. Secondly, compare entry pages within one source to see whether that audience performs better when it lands somewhere else.
Controlled traffic can help with quality assurance before a live campaign. An AI tool for website traffic campaigns can provide planned visits with configurable targeting, device options, pacing and campaign monitoring, which you can use on sites you own or are authorised to test. Label that traffic clearly and use it to validate UTMs, landing-page rows, paths and dashboard filters, not as evidence of genuine demand or commercial performance.
Once tracking is verified, let real visitor outcomes guide optimisation. Increase relevant traffic to a promising page gradually, keep the source and creative clearly labelled, and compare the new cohort with a stable baseline. If efficiency falls as volume rises, the original page may have served a narrow audience rather than possessing unlimited scaling potential.
A practical monthly review looks like this:
| Step | Decision |
|---|---|
| Validate | Confirm revenue, currency, key events, UTMs and landing-page rows |
| Rank | Compare total revenue and revenue per visitor with sample size visible |
| Diagnose | Review key-event rate, depth, source, device and paths |
| Prioritise | Choose one high-volume weakness and one high-efficiency opportunity |
| Test | Change one substantial variable and annotate the launch date |
| Review | Compare equivalent periods and visitor groups |
Avoid shortcuts. Do not send unrelated traffic merely to lift session counts, mark trivial actions as key events, or combine sources until every page appears average. Those tactics make the dashboard calmer while making the decision worse.
Your next step is to export the Landing page report and identify the ten pages with the most sessions. Add revenue per visitor, the key-event rate for one genuine business outcome, Views per session and Session source / medium. By the end of that review, you should know which page deserves more relevant traffic, which page needs repair and which apparent winner is simply benefiting from volume.
Do that before your next campaign launch. When you rank entry pages by value rather than visits, you stop buying attention blindly and start building journeys that work for the real people arriving on your site. That evidence should also shape where your digital marketing focus goes next.
GA4 calls it a landing page. The Landing page dimension records the page path and query string associated with the first page view in a session, which makes it session-scoped. That scope matters, because it should be paired with session-scoped metrics and session-scoped acquisition dimensions rather than user-scoped ones.
Either works, provided you use the same one for every page in the comparison. Revenue per active user estimates the value generated per person represented in the report, while revenue per session measures the value generated per visit. Mixing the two across pages produces a ranking that means nothing.
No. It is revenue associated with sessions that began on that page. Visitors often read several pages, leave and return later, or receive a nudge from another campaign before buying. Treat the ranking as a guide for where to look, not as causal attribution.
Not by itself. Several page views can mean considered research or plain confusion and backtracking. A one-page session can be a fast, successful journey if it produces the key event. Read Views per session alongside engagement time and the selected key event before drawing a conclusion.
Monthly works for most businesses, over a date range covering at least one complete business cycle, and always before a campaign launch. Validate the tracking first, then rank, diagnose, pick one high-volume weakness and one high-efficiency opportunity, and annotate every change so you can connect movement to a real cause.
