Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Experiment Library

Gated Content Test

A test plan for deciding whether a content asset should sit behind a form: gated and ungated arms read on qualified leads per visitor, not form fills.
Experimentadvanced6 min readUpdated

Hypothesis

If a content asset is offered without a form, more visitors will consume it and enough will later convert through other routes that qualified pipeline per visitor matches or beats the gated version, which captures contacts but deters readers. The decision behind it: keep a content asset behind a form, or make it freely available.

The gating argument is usually conducted in principle. One side counts form fills and sees a lead engine; the other counts readers lost at the form and sees an asset nobody reads. The lead generation versus demand generation guide sets out both positions; the B2B content guide explains which assets tend to be worth gating. For a particular asset the answer is empirical, and this test supplies it. Name the asset, the optional sign-up offered in the free arm, the horizon and the acceptable margin in the experiment hypothesis template before launch.

Metrics and guardrails

The primary metric is sales-qualified leads per asset page visitor, at a fixed horizon: of all visitors assigned to an arm, the share who become a Sales-Qualified Lead by any route before the horizon, whether through the asset's form, the optional sign-up or a later demo request, webinar or contact form. The denominator is every assigned visitor. Form fills and marketing-qualified leads are reported for context, but the gated arm wins them by design.

Because qualified leads are rare and late, many teams cannot wait for them alone. If a proxy is needed for an interim view, such as leads reaching a defined engagement stage, name it before launch and keep the final decision on the primary metric. The content goals and KPIs guide explains how consumption metrics relate to business outcomes; it is where the proxy should be agreed.

Three guardrails cover what each arm gives up. Asset consumption rate, the share of visitors who open or read the asset, shows the reach the gate costs. Contacts captured, from the form or the optional sign-up, shows what the free version loses for nurture programmes. Organic visibility of the asset page, read in Google Search Console for the page's clicks and impressions, shows whether the test itself changed how the page performs in search.

Design

Randomise at visitor level, on the first visit to the asset landing page, and keep the assignment for later visits. The gated arm shows the current form before download. The free arm shows the asset directly, with an optional, clearly secondary sign-up for related content. The page copy, design and the asset itself are identical. Store the arm in a first-party identifier that later forms on the site can read, so any lead created afterwards can be joined back to it.

Design elementChoiceReason
Unit of randomisationVisitor to the asset page, assigned on first visitThe gate acts on the visitor's first encounter
ArmsForm before download versus free access with optional sign-upOne change: whether the form is required
AttributionArm stored and read by every later formUngated readers convert through other routes
Primary metricSales-qualified leads per visitor at a fixed horizonCounts every route, not only the asset form
Decision marginLargest acceptable loss written before launchMakes "matches" a testable statement
Search versionOne version served to crawlers throughoutPrevents the test changing organic visibility

The lead form length test is the nearest design: it shortens a form, while this one removes it. If the gate stays, that test is the natural next step.

Sample size reasoning

Qualified leads per visitor is a rare outcome that arrives late, so the sample is large and the horizon long. The inputs are the gated arm's expected qualified leads per visitor, the margin of acceptable loss and the confidence required. The sample size and significance guide explains how these combine, why a rare outcome needs many visitors, and why a test on a rare outcome often reads a proxy, which must be named in advance. Testing whether one arm is no worse than another by a stated margin can need as many visitors as testing for a difference, so do not assume the free arm only has to avoid losing.

If the asset page does not attract enough visitors for the primary metric, run the test on a group of similar assets together, each split the same way, or accept the proxy as the decision metric and say so before launch.

Duration

Enrol until the planned number of visitors is reached, across full weekly cycles, then wait for a horizon long enough for ungated readers to come back through other routes and be qualified. The horizon should reflect how long the business's buyers usually take from first content contact to a sales conversation; the B2B measurement guide covers how that lag is measured. Avoid promoting the asset in a campaign that sends traffic to only one version.

Reading the results and the decision rule

Read once, at the horizon. Compare sales-qualified leads per visitor with its confidence interval against the pre-set margin, then the guardrails. The rule, written before launch:

ResultDecision
Free arm's qualified leads per visitor match the gated arm within the margin, or beat itUngate the asset
Free arm loses more qualified leads than the margin allowsKeep the gate
Free arm matches, but organic visibility or contacts captured fall beyond agreed limitsUngate with a stronger optional sign-up, then recheck
Result too uncertain to judge against the marginKeep the current state and pool more assets into a larger test
Illustrative example — not a real result

A security training firm tests ungating its playbook

Keelson, an invented cybersecurity awareness training company, splits 30,000 visitors to its phishing-response playbook page over ten weeks. The gated arm captures 2,100 form fills; the free arm gets 900 optional sign-ups, and consumption rises from 13% to 58% of visitors. At a 120-day horizon, the gated arm has produced 62 sales-qualified leads and the free arm 70, or 4.1 and 4.7 per 1,000 visitors; the 95% interval for the difference is roughly −0.9 to +2.0 per 1,000. The agreed margin was a loss of no more than 1.0 per 1,000, so the free arm matches the gated one within it. Organic clicks to the page are steady, and Keelson ungates the playbook.

Pitfalls

Sharing of the ungated asset reaching the gated arm is the first pitfall. A free PDF passed around by email or posted in a community reaches people assigned to the gated arm, who then read it without the form. That contaminates the comparison in the gated arm's favour on reach and blurs both arms' outcomes; it cannot be prevented entirely, so state it as a limit and keep the asset's direct file link unlisted during the test. Search engines indexing one version is the second. If crawlers see the gated page sometimes and the free page at other times, the page's search performance may change during the test; serve crawlers one version consistently and watch organic visibility as a guardrail. Reading form fills is the third, and it makes the gate win every time.

Common mistakes

Gating tests are decided by what is counted and how long the team waits. The experiment library lists the other lead handling designs.

  • Counting form fills as the outcome. The gate wins them by construction.
  • Reading before ungated readers return. Set a horizon that matches the buying lag.
  • Losing the arm after the first visit. Every later form must read the stored arm.
  • Leaving "matches" undefined. Write the acceptable margin before launch.
  • Promoting one version in a campaign. Paid traffic must reach both arms through the same split.
  • Letting crawlers see both versions. Serve search engines one version throughout.

Frequently asked questions

Why not judge the test on form fills?

Because the gated arm is the only one that requires a form, so it will always capture more contacts. The question is whether those contacts become more qualified leads than the extra readers of the free version do through other routes. Sales-qualified leads per visitor, counted at a horizon that gives ungated readers time to return, answers that question; form fills cannot.

How can ungated readers be traced if they never fill in a form?

Through the same visitor identifier that assigned them to the arm. If an ungated reader later requests a demo, signs up for a webinar or fills in any other form on the same device, the identifier links that lead back to the arm. Some readers will return on another device or be referred by a colleague and cannot be traced; that undercounts the ungated arm, which is a stated limit of the design.

What does 'matches' mean in the decision rule?

It means the ungated arm is not worse than the gated arm by more than a margin agreed before launch. Write down the largest loss of qualified leads per visitor the team would accept in exchange for wider reach, and read the confidence interval against it. Without a pre-set margin, a small and uncertain difference can be argued either way.

Terms used on this page

Frameworks

Also related

Related