TL;DR abstract. The most repeated rule in paid search landing pages is never to send ad traffic to your homepage. Doctor Ads Profit Forensics compared homepage clicks with deep-page clicks inside each account, so the advertiser and the year are the same on both sides. The traffic is not: brand searches, sitelinks and different products can land on each side, and the landing page report cannot separate them. In 2025 the median account’s homepage converted at 0.96 times the rate of its own deeper pages. It converted worse in just over half of the accounts and better in the rest, and the same comparison gave 0.98 in 2023 and 1.02 in 2024.
The median hides the real finding. The result is almost never a tie. In a quarter of accounts the homepage converted at less than half the deep-page rate, and in another quarter at 1.77 times or more. The median account sent 25.9% of its landing-page spend to the homepage, and accounts split into two camps: under a tenth, or over half. Separately, pages that never recorded a conversion were almost half of all pages that received spend but held only 3.0% of the money. The rule “never the homepage” does not describe these accounts in either direction: in some accounts the homepage converted well below the rest of the site, in others well above it.
Ask PPC practitioners where a paid click should land and most will give the same answer: anywhere but the homepage. Unbounce put it in one line in its message match audit: “If you point your ad at your website’s homepage you’re doing it wrong.” The advice is old, it is intuitive, and it is almost never measured. What gets published instead is a survey of what marketers say they do, a pooled benchmark of landing page builders, or an agency comparison that sets one of those against the other.
Here is what the published record actually holds. MarketingSherpa’s Landing Page Handbook reports a 2007 survey in which about half of marketers said their ad, email and search clicks went to a homepage rather than a dedicated page, plus a small observational sample of 920 promotional links in which 26% landed on a homepage. Unbounce’s conversion benchmark puts the median landing page built on its platform at 6.6%, across 41,000 pages, which says nothing about homepages because none were measured.
The largest published study is academic, and it points the other way. In “What happens after an ad click?” (CIKM 2009), researchers at Yahoo classified the landing pages of sponsored search ads and found that advertisers who chose the homepage tended to have higher relative conversion rates than the dataset average. The authors were careful to say this does not prove anyone chose the right page. Their comparison pooled many advertisers together, so a homepage from one business was compared with a category page from another.
Same-advertiser comparisons do exist as single tests. Instapage split its own branded search traffic between its homepage and a dedicated page and reported that the dedicated page won; the two ads also carried different headlines. What we could not find is the distribution: across many accounts and several years, how the homepage compares with the rest of the same site. This is an observational comparison of the traffic each account actually routed to each kind of page, not a controlled test of the page type.
The median account sends a quarter of its landing-page money to the homepage
Start with how much traffic the rule is even about. In 2025 the median account sent 25.9% of its landing-page spend to its homepage. The middle half of accounts ran from 5.7% to 77.1%, which is less a range than a split. About four accounts in ten sent under a tenth of their money to the homepage. Another four in ten sent more than half. The middle is thin.
The first half of 2026 gives a median of 34.9%, but that is a different mix of accounts, not a trend. Accounts measured in both 2024 and the first half of 2026 moved by a median of about one percentage point, and somewhat more of them rose than fell.
One number we deliberately do not lead with: pooled across all landing-page spend, the homepage share in 2025 was 23.3%. A single large account carried more than half of that window’s spend and pulled the pooled figure down. Without it, the pooled share is 40.6%. When one account can swing a pooled number by seventeen points, the median account is the honest headline.
Inside the same account, the homepage converts at 0.96 times the deep-page rate
For the test itself we kept only accounts that sent at least 100 clicks to the homepage and at least 100 clicks to deeper pages in the same year, and that recorded at least one conversion on the deeper pages. That excludes one-page sites, where the homepage is the landing page, and accounts that never used the homepage at all.
For each of those accounts we divided the homepage conversion rate by the deep-page conversion rate. A value of 1.0 means the homepage converts exactly like the rest of the site. Conversions are each account’s own primary conversions, so they are never added up across accounts. Only the ratio travels.
The median ratio in 2025 was 0.96. Leave out any single account and it moves between 0.92 and 0.98: the median sits in a thin part of the distribution, where few accounts are near parity, so one account can shift it by several hundredths. The homepage converted worse in just over half of the accounts and better in the rest. It lost a little more often than it won.
It is almost never a tie
A median near 1.0 invites the wrong reading, that homepages are as good as landing pages. The distribution says something sharper. In a quarter of the accounts the homepage converted at less than 0.41 times the deep-page rate, less than half as well. In another quarter it converted at 1.77 times or more, and a quarter of all accounts reached twice the deep-page rate. The homepage was at least 20% worse in about as many accounts as it was at least 20% better.
Accounts within 20% of parity were the rare case, fewer than one in six. Raise the bar to 300 clicks on each side, or to ten conversions on each side, and the share of near-ties stays about the same while the median moves to 0.98 and 1.06, though the middle half narrows to 0.73 to 2.06 on the stricter filter. Part of the spread is sampling noise and part is different traffic on each side; this analysis does not separate the two for any single account.
So the useful sentence is not “homepages are fine”. It is this: in most of these accounts the observed homepage rate sits well away from the rest of the site, in one direction or the other, and a general rule cannot tell you which way yours leans.
Three years, the same coin
A single year could be an accident. The same comparison on 2024 data gives a median of 1.02, with the homepage losing in half the accounts. On 2023 data it gives 0.98, losing in just over half. Different accounts enter each year, so this is a repeat, not a trend line, but three separate years land within a few hundredths of 1.0.
The first half of 2026 gives 1.19, but we do not use it. Conversions from May and June clicks were still arriving when the data was pulled, and the recent months fill in unevenly. It is listed in the method section as a sensitivity check, not as a finding.
The clicks cost about the same
If homepage clicks were much cheaper, a lower conversion rate could still pay. In 2025 they were not: the median ratio of homepage cost per click to deep-page cost per click was 0.99. We do not generalise this one. In 2024 the same ratio was 0.61 and in 2023 it was 0.92, so the price of a homepage click moves more from year to year than its conversion rate does. The conversion comparison is the stable part of this study; the price comparison is not.
Is it just brand traffic?
The strongest objection is brand. People who search for your company name tend to land on the homepage and tend to convert well, so a homepage that looks competitive might simply be borrowing the strength of branded searches. The landing page report does not say which keyword or campaign sent each click, so we cannot split brand from non-brand directly.
One exploratory check is whether the ratio moves with how much of an account’s money goes to the homepage. If a sliver of homepage traffic were mostly brand, those accounts might show a flattering ratio, and accounts routing most of their non-brand spend there a poorer one. That is only one way brand could show up; homepage share is not the brand share of homepage clicks.
Split the accounts into thirds by homepage share and the median ratios come out at 0.86, 1.28 and 0.77. That is not a slope in either direction, and the rank correlation between homepage share and the ratio is minus 0.02, which is close to none. This check cannot rule brand in or out: brand traffic could shape the ratio in ways that do not track homepage share at all. It stays the largest open question behind this result.
In the median account, three pages carry 80% of landing-page spend
The homepage debate assumes a site with many candidate pages. Most accounts do not spread their money that way. In the first half of 2026 the median account paid for clicks on 21 distinct landing-page URLs, but its single top page took 51.3% of the spend and its top three took 85.3%.
In the median account, three pages are enough to reach 80% of landing-page spend. In about one account in four, a single page carries 80% or more on its own. The 2025 numbers repeat the shape: a top page at 46.8% and three pages to reach 80%.
That changes where attention belongs. The question is rarely “homepage or landing page” across hundreds of URLs. It is whether the two or three pages that take most of the budget are the right pages, and whether each of them does its job. For many accounts one of those pages is the homepage.
Pages that never convert: half the pages, 3% of the money
The last test looks for waste at the page level. In 2025, landing pages that recorded no conversion at all made up a median 46.4% of the pages that received spend in an account. They held a median 3.0% of the money. Leave out any one account and the median stays between 2.9% and 3.1%; pooled across all accounts it is 1.5%. The 2024 data gives 3.7%.
Read this with its unit in mind. A page counts as converting if any click on it converted, so a page with one conversion and a lot of unprofitable traffic sits entirely outside the 3.0%. The figure says that spend rarely lands on pages with no recorded conversion at all; it does not say how much of the spend on converting pages paid back, and it cannot be set against search-term waste, which is counted query by query. Our earlier study of wasted spend in search terms measures a different thing on a different unit.
When this does not apply
Your homepage is your only page. A one-page site has no deep pages to compare against, and these accounts were excluded from the test by design.
You sell many products and your ads point to each one. The comparison needs clicks on both sides. If almost all of your traffic goes to product pages, the question barely arises, and your two or three biggest pages matter more than the homepage.
You changed the site during the year. The report groups clicks by URL, not by what the page looked like. A redesign in the middle of a year blends two different homepages into one row.
You compare with a benchmark instead of with yourself. The 6.6% median landing page conversion rate from Unbounce and the account averages that WordStream publishes are levels from other businesses. A homepage converting at 3% is only bad if your own deeper pages do better. That comparison is available in your own account and it is the only one that holds the business constant.
Your account has never been through an audit. This is a managed portfolio, not a random sample of advertisers. Accounts that route traffic to the homepage here have usually been looked at by someone, which is not the same as the average advertiser who never chose a final URL at all.
How to run the test on your own account
- Open the Landing pages report for the account and set the date range to a full year, or at least a full quarter that is more than a month old, so recent conversions have arrived.
- Export it and mark every row whose URL, with tracking parameters removed, is your homepage or a language root such as /en/.
- Add up clicks and conversions for the homepage rows and for everything else, and divide one conversion rate by the other. On thin volumes a ratio far from 1.0 is a question, not a verdict; check which searches and sitelinks send clicks to each side before you act.
- Do not stop at the ratio. Sort the other rows by spend and look at the top three. In most accounts they carry most of the money, and they are where a page change pays back first.
- If the homepage wins, find out why before you move traffic. Brand searches may be doing part of the work.
Key takeaways
- Inside the same account, the homepage converted at a median 0.96 times the deep-page rate in 2025, and at 0.98 and 1.02 in 2023 and 2024.
- The homepage lost in just over half of accounts and won in the rest; near-ties within 20% were fewer than one account in six.
- In a quarter of accounts the homepage converted at under half the deep-page rate, in another quarter at 1.77 times or more: observed ratios fall well away from parity in both directions.
- The median account sent 25.9% of landing-page spend to its homepage, and accounts split into under a tenth or over half.
- Three pages carried 80% of landing-page spend in the median account; pages that never converted were almost half of all pages and 3.0% of the money.
Researcher’s take
I have said for years that when ads underperform, the problem is usually not the ad but the page it lands on. This data does not contradict that, and it cannot prove it either. My reading, which the numbers here do not test, is that what decides is not whether the page is called the homepage but whether that particular page answers the search that paid for the click. I have seen homepages that convert better than any landing page an agency built, and I have seen homepages that quietly cost an account half its conversions. The rule was a shortcut for people who did not have time to look. You have the data to look, and it takes an afternoon.
Igor Ivitskiy, Doctor Ads · 19 years · $770M managed spend
Method
Doctor Ads Profit Forensics is the research desk of a Google Ads practice with 19 years and $770M managed spend behind it.
Data statement v.2026.09. Source: the landing page report (unexpanded final URL level) of a managed Google Ads portfolio analysed by Doctor Ads Profit Forensics, read from a snapshot dated 10 July 2026 with yearly windows 2023, 2024 and 2025 and a half-year window for January to June 2026. Accounts are pseudonymised; no account, domain, URL or vertical is published.
Definitions. URLs are normalised by removing tracking templates, the whole query string and fragments, and lowercasing the URL, so a landing page here is a cluster of URLs that differ only by parameters or letter case. The language-root rule for the homepage is a heuristic. The homepage is the root path, /index.html, /index.php, /home, or a bare language root: a two-letter ISO 639-1 code such as /en/ or /en-gb/, plus /ua/, which Ukrainian sites use for their Ukrainian version. Two-letter paths that are not language codes, such as /rf, count as deep pages. Every other parsed URL is a deep page. 63 rows whose URL had no parsable host, unfilled templates carrying $18,605 across four windows, are excluded. Share figures use accounts with at least $500 of landing-page spend in the window: 74 accounts and $7,171,786 in 2025, 57 accounts and $4,278,060 in the first half of 2026, 58 in 2024 and 47 in 2023. The within-account comparison uses accounts with at least 100 homepage clicks, at least 100 deep-page clicks and at least one deep-page conversion in the window: 53 accounts in 2025 (2,200,123 homepage clicks against 5,853,601 deep-page clicks), 42 in 2024 and 39 in 2023. Concentration uses 57 accounts in the first half of 2026 and 74 in 2025. The zero-conversion page test uses the 67 accounts with at least one conversion in 2025.
Weighting. Every headline figure is a median across accounts with the middle half reported, and each is recomputed leaving one account out at a time. Conversion rates are compared only inside the account that recorded them. Pooled shares are shown only next to their value without the largest account, which held 54.0% of 2025 landing-page spend. Medians and quartiles use one engine with linear interpolation.
Robustness. Root path only as the homepage: median ratio 0.89. At least 300 clicks on each side: 0.98. At least ten conversions on each side: 1.06, with 7 of 43 accounts within 20% of parity. Counting non-language two-letter paths such as /rf as the homepage, as a first version of this analysis did, gives 0.98 on 52 accounts. First half of 2026, with immature conversions: 1.19. Brand check: thirds by homepage share give 0.86, 1.28 and 0.77, rank correlation (average ranks) minus 0.02.
Limitations. Five. The landing page report does not carry campaign or keyword, so branded and non-branded traffic cannot be separated. Google’s documentation says the report covers final URLs across Search, Display, Video and Shopping and, for Search, sitelink URLs, so a homepage row can include sitelink clicks. Conversions follow each account’s own settings and attribution. Nothing here is causal: advertisers chose where to send traffic, the same page type means different things on different sites, and 24 of the 53 compared accounts spent on more than one host. And the portfolio is agency managed, not a random sample of advertisers.
Changelog. First published September 2026. Recomputed quarterly against the live claim registry.
What to read next
- Which Google Ads score actually tracks results, including the landing page speed test in the same portfolio.
- How much of your Google Ads budget is wasted, on waste measured query by query, a different unit from the page-level figures above.
- The pooled prior: Becker and colleagues, “What happens after an ad click?”, CIKM 2009.
