How to Label Free Articles on a Paywalled Site: Test Plan

Your live blog is free. Your explainer is free. Your breaking-news page is free. But a reader scanning a search result or your homepage has no way to know that, and some will assume a paywall and skip it.

The New York Times takes the opposite approach. According to Search Engine Journal's analysis, it gives prime homepage space to free articles and makes sure readers know they are free. The author argues that labeling improves engagement and reach. That is one writer's opinion, not a tested result, so this guide turns it into an audit and an experiment you can run on your own site.

Step 1: Inventory the free articles you already have

Start with a list, not a design. Free status can live in several places at once, so you may not be able to say which URLs are free on a given day.

1. Export every article from your CMS where the paywall rule is set to free, plus any with a start or end date on that rule.
2. Pull articles that bypass the meter through other routes: live blogs, breaking-news templates, syndicated or wire pieces, public-interest explainers, and campaign landing pages.
3. Cross-check against a logged-out crawl. Fetch a sample of URLs as an anonymous visitor and record which ones render full text. The CMS flag and the real experience sometimes disagree.
4. Record the expiry. If an article is free for 48 hours and then metered, any label needs a removal rule.

The output is one sheet with URL, template, free reason, start date, end date, and verified free (yes or no).

Step 2: Classify each article's traffic role

A label only matters where someone is deciding whether to click. Join your inventory to analytics and Search Console data, then tag each URL by how people reach it.

The source article names live blogs as a prime case. Many subscription publishers make them free because they compete hard in Search and bring in new audiences, yet rarely say so. Treat that as a hypothesis to check against your own data, not a fact about your site.

Useful tags include:

• Search entry point (high impressions in Search Console)
• Homepage and section-front traffic
• Newsletter and social referral
• Low-visibility free content that few people see

Also note the audience: mostly anonymous, registered, or subscribed. Labels mainly address the anonymous group.

Step 3: Audit current labeling and paywall markup

For each high-priority URL, record where "free" is already signaled, if anywhere.

1. Homepage and section-front modules: is there a free block, and does it say so in words?
2. Card badges: do teasers carry a text label, or only an icon or color?
3. Featured image: is any badge baked into the image file?
4. Open Graph image and social metadata: is a label present, and is it consistent across templates?
5. Structured data: view the rendered page source and check how paywalled-content markup is implemented.

Do that last check before you add anything. Google's paywalled-content structured data uses properties such as isAccessibleForFree to describe which parts of a page are gated, and it exists so paywalls are not mistaken for cloaking. Confirm that your free articles are not accidentally marked as gated, and that gated articles are not marked free.

The source article suggests experimenting with free badges "in the structured data", but it does not say which property would carry a visible label. It also does not say whether any markup produces a visible badge in Search. Until Google's documentation says otherwise, treat markup as a correctness task, not a labeling tactic.

Step 4: Prioritize what to label first

Score each free URL on three questions. Does it get many impressions or much homepage exposure? Is anonymous traffic likely to see it? Does it stay free long enough to justify the work? For more on this, see more on how to build an html sitemap that helps users and seo.

Start with the intersection: high-exposure, anonymous-heavy, stably free content, such as a recurring live-blog template or evergreen explainers. Skip short-lived free items at first. A label that outlives the free window can create a trust problem worse than the one you are solving.

Step 5: Choose where the signal goes

Each placement has a different cost and a different risk.

Homepage module. The source says the NYT's most trafficked section is free, and it ties that to a goal of eventually funneling readers to the app. Decide your equivalent goal first. Risk: free content competes with subscription prompts for the same space.

Card badge. This is the cheapest and easiest placement to test. Use a text label ("Free to read"), not color alone, so it works for accessibility. Risk: badge clutter if too many items carry it.

Featured image. The source suggests badges on the featured image. Risk: the label is baked into pixels, so it is hard to remove when free status ends, and it may crop badly across surfaces. We cover related ground in go deeper on growth leader first 30 days: find vs believe channel audit.

OG image. The source notes that OG images are applied less consistently on third-party platforms. Risk: platforms cache previews, so changes and removals may lag, and you cannot control how they render.

Structured data. Correctness comes first, as above. Do not treat it as a display label.

Step 6: Design the test

Pick one success metric before you start. For homepage and section-front placements, use click-through rate on the labeled card among logged-out visitors. Pair it with guardrail metrics so a click gain that hurts the business does not read as a win.

The cleanest design is a randomized A/B split on-site: half of logged-out visitors see the card with a label, half see it without. You cannot randomize Search surfaces this way. For featured-image and OG-image changes, use a before/after comparison, with a matched set of similar free articles left unchanged as a control.

Worked example (illustrative only)

This is a hypothetical mid-size news site, "Example Daily." It is not a real publisher, and the table shows no results because none exist.

Example Daily finds that its recurring live-blog template and a set of evergreen explainers are free, verified in a logged-out crawl. It adds a text badge reading "Free to read" to homepage cards for those items and splits logged-out visitors 50/50 for two weeks. Separately, it adds the badge to the featured image on half of a matched set of explainers and leaves the rest as controls.

| Metric | Control | Variant | How to read it | |---|---|---|---| | Card CTR (logged-out, homepage) | To be recorded | To be recorded | Primary metric; compare within the same time window | | Scroll depth on the article | To be recorded | To be recorded | Guardrail; a CTR rise with falling depth may mean the label draws low-intent clicks | | Returning visits within 14 days | To be recorded | To be recorded | Closer to subscription intent than a single session | | Registrations | To be recorded | To be recorded | May be too sparse for a two-week test; treat as directional |

Also read: related topic: how to prepare shorts for ask youtube: a creator audit

Step 7: Pick metrics that might predict subscriptions

The source asks whether session duration or returning visits predict a future subscription, and says each brand must establish what matters to it. It offers no answer, and neither does this guide. Test it internally: look at past subscribers and check whether they showed repeat visits or registrations before converting. Only then promote a metric to your decision rule.

Step 8: Verify and know the limits

Run this checklist before and after the test.

1. In analytics, confirm that event tracking fires identically on control and variant, and that logged-in users are excluded from the primary metric.
2. In Search Console, compare impressions, clicks, and CTR for the affected pages against the control set. Annotate the change date.
3. Validate structured data on a sample of labeled and unlabeled URLs with Google's testing tools, and confirm that paywall markup matches reality.
4. Check the OG preview in the sharing tools of the platforms you care about, remembering that cached previews can lag.
5. Rule out confounders: a major news event, a homepage redesign, a newsletter send, an algorithm update, or a seasonal swing during the test window.

Now the limits. News-cycle noise can swamp a small effect, and live blogs spike unpredictably. Small samples, especially for registrations, mean results may be directional at best. Isolating SEO effects is hard because a featured-image change alters several things at once.

The source also says engagement is "almost certainly" Google's most important ranking factor. The article offers no evidence for that claim, and this guide does not rely on it. A label may change CTR and sessions without moving rankings, so measure clicks and behavior, not position, and do not promise rankings to stakeholders.

If the test window overlaps a big news event or the sample is thin, do not draw conclusions. Extend the test or rerun it.

What to expect next

A first round may produce a cleaned-up inventory, a few fixed markup mismatches, and an ambiguous or modest CTR read. Use that result to decide whether a second, longer test on a narrower set of stable free content is worth the effort. Keep the removal rules in place so labels never outlast the free window.