How to test long-form copy against a short landing page

The long landing page sat under the ads for months. Last Tuesday someone published a short version in its place. Form fills ticked up, paid spend also rose, and the team called short copy the winner.

That is a before-and-after swap, not an A/B test. A valid length test splits the same traffic, changes only how much copy sits with the form, and names the conversion before launch.

Set the split so Google does not treat the variants as cloaking. Leave the test up only until the tool can pick a winner, then ship that version.

Change only how much copy sits on the page

Keep the offer, price, headline job, and form fields identical. The long variant adds proof, FAQ, and objection copy above or beside that same form. The short variant keeps a short benefit list and stops.

HubSpot's landing page A/B guidance lists copy length as one variable to test. It also lists form fields, media, and calls to action as separate variables. Mixing two of those in one test means you cannot tell which change moved the form.

A pricing table that overflowed on phones is a layout problem. Fix overflow first, or the short variant will look like it won because people could finally tap Choose plan.

These setups are not a length test. They mix audiences, offers, or metrics that move for reasons other than how much copy you published.

• Swap the live page mid-week and compare this week to last week.
• Send paid clicks to a short URL and organic visits to the long URL.
• Change the form fields in the same test as copy length.
• Use bounce rate as the primary metric because short pages load faster.
• Stop after a few dozen visits because one variant looks better.
• Show Googlebot only the original while humans see the short page.
• Use a 301 redirect from the original URL onto the variant.

Name the conversion in writing before anyone publishes B. For most B2B landing pages that is a form submit or a booked call, not time on page.

Split the same audience without cloaking

Both variants need a fair share of the same mix of paid, email, and organic visitors. Sticky assignment inside a session stops the same person from seeing long copy, then short copy, on refresh.

Same-URL test in HubSpot

HubSpot A/B tests two versions of a page at the same live URL. Half of visitors see one version, half see the other. Preview URLs differ so you can check B without changing the public path.

Open Content, then Landing Pages, and click the page. In the left sidebar, choose the Test icon, then A/B test, and name both variations. Edit B so the body is shorter and nothing else moves.

Publish now so both versions go live. HubSpot keeps the same variant inside one session. After the session ends, the next visit can land on either version.

That feature sits on Content Hub Professional and Enterprise. Without it, do not fake a split by swapping templates by day of week. Use extra URLs and a temporary redirect, or wait until the CMS can split traffic.

HubSpot's A/B test help is the menu path. Review results on Performance, then the Test results tab. Choose as winner publishes the winner and reverts the loser to draft.

Extra URLs with a temporary redirect

Some stacks cannot serve two bodies on one path. Google Search Central's website testing guidance covers that case. Redirect some users from the original URL to a variant URL with a 302, not a 301.

A 301 tells Google the move is permanent. The variant can replace the original in the index. A 302, or a JavaScript redirect, marks the split as temporary while the experiment runs.

Put rel=canonical on every variant URL back to the original. Google prefers that over noindex on the test URLs. Noindex on a homepage-style test can have unexpected effects, including dropping the URL you still want indexed.

Do not show Googlebot a different set of URLs than humans. That is cloaking, whether you do it in server logic or robots.txt. Googlebot generally does not keep cookies, so it sees the version that loads without them.

Leave the no-cookie version as the current original, usually the long page you already rank with. Do not write a Googlebot exception that always serves the short page. The crawler should get the same default a cookieless browser gets.

Call the winner on the conversion you named

Short pages often show less time on page and a different bounce pattern. That is expected when there is less to read. It is not proof the short page lost, or won, the form.

Read form submits, or booked calls, against the traffic each variant actually received. Segment paid from organic if bids or budgets moved during the window. A spend increase on the short URL is not a copy win.

Google's testing notes say a good tool tells you when you have enough data for a reliable conclusion. HubSpot exposes that comparison on Test results. If volume is too low for the tool to separate the variants, do not pick a winner by gut.

Once you conclude the test, ship the winning copy on the original URL. Remove variant URLs, test scripts, and leftover redirects as soon as you can. Google may treat an experiment that runs far longer than needed, especially one that serves one variant to most users, as an attempt to deceive Search.

Unlike a full CRO and experimentation engagement, this post covers only how to isolate page length. For a look at whether the current long versus short split is a real test or a Tuesday swap, start with Get Free Assessment.