How to A/B Test Your Newsletter Subscription Price in 2026
The median beehiiv newsletter converts 0.62% of free readers. Here's how to design a price test that actually finishes, and read the winner in revenue.
·
The short answer: Run two prices at a time, not four, and score the winner in revenue per paywall visitor rather than conversion rate. At beehiiv's median free-to-paid rate of 0.62%, a four-price test needs roughly 241,600 paywall visitors to detect a 25% difference, which is about 20 years of traffic for most newsletters. Two arms, a materially different price, and revenue as the goal is the only version that finishes. See revenue A/B testing and the sample-size calculator.
Every guide on pricing a paid newsletter gives you a framework for guessing. Answer four questions, consider your costs, think about the value you deliver, pick a number.
Then you pick the number and never touch it again.
The data says that number is probably wrong. Across thousands of publications, beehiiv's State of Paid Newsletters 2026 found a median price of $10 per month and $100 per year, unchanged since 2024. A price that hasn't moved in two years across an entire platform is not a market equilibrium. It's a default.
This guide is the other half of the pricing conversation: how to measure the right price instead of reasoning your way to one. It covers what newsletters actually charge in 2026, the sample-size math that decides how many prices you can test at once (this is where most price tests quietly fail), what to measure, what real price tests have found, and the exact setup on beehiiv.
How much should you charge for a newsletter subscription?
Start at $10 per month and $100 per year for a general-interest newsletter, then adjust for your subject: $15 for business, $20 for finance, $27 for investing, $7 for food and travel. Those are beehiiv's 2026 medians, and the median overall has not moved since 2024. The median hides most of the useful signal, though, because price varies far more by subject than by list size. Investing newsletters run a median of $27 per month. Food and travel newsletters run $7. A nearly 4x spread.
Here's the full spread from beehiiv's 2026 analysis:
| Niche | Median monthly | Median annual |
|---|---|---|
| Investing | $27 | $292 |
| Finance | $20 | $200 |
| Business | $15 | $150 |
| Marketing / Technology / News / AI | $10 | $100 |
| Sports | $8 | $80 |
| Entertainment | $8 | $80 |
| Travel | $7 | $80 |
| Food and drink | $7 | $70 |
Two things worth pulling out of that table.
First, the range is nearly 4x from food to investing. If you are charging $10 because $10 is "the newsletter price," and you write about investing, you are likely leaving more than half your revenue on the table.
Second, these are medians of what operators chose, not of what audiences will bear. Ghost's guide to pricing a subscription newsletter notes most newsletters land in the $5-15 per month band and asks the obvious question: why? There's no rule that says those are the boundaries.
Jessica Lessin, founder of The Information, put it more bluntly to the founder of ImpactAlpha, who had researched his way to $100 per year. Her reply: "One of the general rules of pricing is that people price too low. I would take it to that point that makes you slightly uncomfortable." She suggested $599. They launched at $399, four times the founder's own research-backed instinct, and grew to a team of 15+ serving 70,000 readers.
That's an anecdote, not evidence. Which is exactly the problem, and the reason to test.
How many prices can you actually test at once?
Two. For nearly every newsletter, the answer is two. This is the section that decides whether your price test produces an answer or produces noise, and almost no pricing guide covers it.
The reason is the conversion rate. beehiiv's 2026 report puts the median free-to-paid conversion at 0.62%, roughly six paid subscribers per thousand free readers (for what moves that number, see free-to-paid newsletter conversion rates). Price tests don't run on your subscriber list, they run on the people who actually reach your upgrade page, and at that conversion rate each variant collects purchases very slowly.
Here's what a four-arm test (one control, three price treatments) requires at 80% power, 5% familywise error, two-sided, Bonferroni-corrected across the three comparisons:
| Baseline purchase rate | Effect you want to detect | Visitors needed per variant | Total across 4 variants | At 1,000 paywall visitors/month |
|---|---|---|---|---|
| 0.62% (beehiiv median) | +25% | ~60,400 | ~241,600 | ~242 months |
| 0.62% | +50% | ~16,800 | ~67,200 | ~67 months |
| 0.62% | +100% | ~5,020 | ~20,080 | ~20 months |
| 3.0% (strong newsletter) | +25% | ~12,150 | ~48,600 | ~49 months |
| 3.0% | +50% | ~3,360 | ~13,440 | ~14 months |
| 3.0% | +100% | ~1,000 | ~4,000 | ~4 months |
Read the top row again. Four price points, a median newsletter, and a 25% effect: twenty years.
Even the friendliest row on the table, a strong 3% converter hunting a doubling, takes four months on four arms. And a doubling is not what price tests find.
The mechanism is simple enough to feel. At 0.62%, four arms splitting 1,000 monthly visitors gives each arm 250 visitors and about 1.5 purchases per arm per month. A variant that "wins" on two purchases versus one has told you nothing. Splitting the same traffic two ways doesn't just halve the wait, it roughly quarters it, because the correction for multiple comparisons goes away too.
So: two prices, materially apart. Then promote the winner and test again against a third. Sequential pairs beat a simultaneous four-way for any newsletter without enterprise-scale paywall traffic. Run your own numbers through the sample-size calculator before you launch anything, using paywall visitors as the denominator, not subscribers.
Should you measure conversion rate or revenue?
Revenue per paywall visitor. Conversion rate alone will actively mislead you on a price test, because the cheaper price wins on conversion almost by definition. What you need is price multiplied by conversion, which is revenue per visitor:
Revenue per visitor = price x purchase conversion
That formula gives you a decision rule you can apply before the test even finishes. A higher price wins on first-payment revenue when conversion holds up better than the inverse of the price increase.
Moving from $10 to $15 is a 50% price rise, so the inverse ratio is 1 / 1.5 = 66.7%. If the $15 arm retains more than 66.7% of the $10 arm's conversion rate, it makes more money per visitor. Below that, it doesn't. Same logic at $10 to $14: conversion needs to hold above 71.4%.
Two caveats that matter more than they sound.
Annual and monthly purchases aren't the same event. A test where one arm sells more annuals looks better on first-payment revenue and may or may not be better over a year. Decide up front whether you're testing a price level or an annual discount, and never move both at once.
Cheap subscribers churn harder. One operator who ran €5, €8, and €15 sequentially reported 9% monthly churn at €5 against 4.7% at €8 (Relijournal). The €5 price converted best (4.1% vs 3.4%) and produced the worst lifetime value, €55 against €170. First-payment revenue and 12-month revenue can point in opposite directions. Re-check the cohorts at 30, 90, and 365 days before you call it permanently.
What have real price tests actually found?
There is no universal direction. Raising price wins sometimes, cutting it wins other times, and the third-party evidence splits cleanly enough to prove the point. Three documented tests, three different answers:
Lower price won. A three-way concurrent test on a paid subscription site ran $10, $12.50, and $14.95 with roughly 8,978 clicks per arm. Orders came in at 156, 94, and 74; revenue at $1,560, $1,175, and $1,106. The $10 price produced 32.8% more revenue than the next best (MarketingExperiments). Demand was elastic, and cutting price made money.
Raising price won, but not where they expected. A subscription business tested a monthly plan at $25 against $33 and $41, leaving quarterly and annual untouched. Monthly purchases fell 17% and 27%. Monthly revenue fell too, 8% and 7%. But quarterly purchases rose 33% and 67%, and total revenue per visitor rose 16% at 99% significance (Conversion). Raising one plan's price pushed buyers toward another plan. If they'd measured only the plan they changed, they'd have called it a loss.
A real newsletter raised and held. The Charlotte Ledger moved from $9 to $12 monthly and $99 to $129 annual in February 2025 with 5,000+ paid subscribers. Five-month retention came in at 92%, new-subscriber acquisition stayed roughly flat year over year, and revenue rose 27%, 25%, and 45% across March, April, and May (Project C). Worth naming the limitation: that's a before-and-after on an existing base, not a randomized test, so list growth and seasonality are baked into those numbers.
The honest read across all three: your elasticity is yours. Nobody's benchmark tells you whether to go up or down. That's the whole argument for measuring it.
Humblytics runs page-level split tests with revenue as the goal and resolves the winner against real Stripe payments rather than click proxies, on a cookie-free 36 KB script with no consent banner. Plans start at $19/month. Start a free trial.
How do you set up a price test on beehiiv?
beehiiv's built-in A/B testing only covers email subject lines, so the price test happens on the web, on your upgrade page. The setup is one tier per price, one upgrade page per tier, and a page-level experiment splitting traffic between them.
Here's the full walkthrough:
Step 1. Create a tier per price. In beehiiv, go to Monetize > Paid Subscriptions and add a tier for each price. New tier, name it, recurring subscription, enter the price, create.
Step 2. Build an upgrade page per price. Duplicate your upgrade page in the beehiiv website builder, then in the copy click the pricing block > Tiers and hide every tier except the one that page is for. Publish. You'll end up with /upgrade and /upgrade-2.
Step 3. Add the tracking snippet. Page-level revenue tests need beehiiv's subscription tracking code alongside the standard Humblytics script so a completed subscription attributes back to the variant that produced it. In Google Tag Manager: new tag, Custom HTML, paste the snippet from the Humblytics docs, trigger on Initialization - All Pages, publish.
Step 4. Create the experiment. In Humblytics, Experiments > Create experiment. Goal: Revenue. Test type: page-level. Control is /upgrade, variant B is /upgrade-2.
Step 5. Turn visitor overlap off. Non-negotiable. Covered below.
Note that the walkthrough above demonstrates four price points because it's showing the mechanics of adding variants. As the sample-size table shows, four is the wrong number for almost everyone. Build it with two.
For testing the headline, CTA, social proof, and layout on that same page, the companion guide is how to A/B test beehiiv newsletter upgrade pages. If your paywall itself needs work first, start with the beehiiv paywall setup guide.
What breaks a newsletter price test?
Four things, and the first one invalidates the whole test rather than just weakening it.
Visitor overlap. If the same reader can see a different price on a refresh or in a later session, your data is unusable and the experience is worse than not testing. Persist the assignment server-side. Select no visitor overlap.
Changing more than the price. If the higher-priced page also has a stronger headline or an added benefit, you measured a package change, not price elasticity. Same copy, same benefits, same trial, same checkout. One variable.
Not grandfathering existing subscribers. Test on new visitors and leave current subscribers on the price they signed up at. Re-pricing an existing base is a trust and billing problem, not an experiment.
Calling it early. Compute the sample size before launch and stop on that number, not on the day one arm pulls ahead. At one or two purchases per arm per week, an early lead is noise with near-certainty. The significance calculator will tell you whether a gap is real.
Frequently asked questions
How much should I charge for a paid newsletter? Start at $10 per month and $100 per year for a general-interest newsletter, which is beehiiv's 2026 median. Go to $15-$25 for business or professional audiences and $20-$30+ for finance and investing, where the reader can attribute direct economic value. Then test upward from there.
How much should I charge for a monthly newsletter versus an annual plan? Price annual at roughly ten months of the monthly rate. beehiiv's standard structure is $10 monthly against $100 annual, which is a 16.7% discount, or two months free. Annual gives you cash up front and removes eleven separate cancellation moments; monthly keeps the entry point cheap. Offer both, and don't test a price level and a discount at the same time.
Can I A/B test pricing inside beehiiv? No. beehiiv's native A/B testing covers email subject lines only. Price testing requires separate upgrade pages plus an external split-testing tool that can split traffic between URLs and attribute revenue back to each variant.
How long should a newsletter price test run? Long enough to hit the sample size you calculated before launch, and at minimum through complete weekly and monthly cycles. At the beehiiv median conversion of 0.62%, two arms hunting a 50% effect need roughly 16,800 paywall visitors per arm. Calculate it against your own paywall traffic, not your subscriber count.
Should I test the annual price or the monthly price? One at a time. beehiiv's standard structure prices annual at $100 against $120 for twelve months of monthly, a 16.7% discount, or two months free. Testing a price level and an annual discount simultaneously leaves you unable to say which one moved the result.
Does a higher price mean fewer subscribers but more revenue? Not reliably. It depends entirely on your elasticity. A $10 to $15 move needs conversion to hold above 66.7% of its former level to make more money per visitor. Documented tests have gone both directions, so the only way to know is to measure your own.
What if my newsletter is too small to test? Then run it sequentially. Hold one price for 6-12 weeks, change only the price, and compare cohorts while controlling for acquisition source and seasonality. It's weaker than randomizing, but it beats four underpowered arms that never reach a conclusion.
Pricing is the one lever that changes revenue without needing a single new subscriber. It's also the lever most newsletter operators set once, on instinct, and never revisit. The fix is not a better framework for guessing. It's a two-arm test, scored in revenue, that you actually finish.
Start a free Humblytics trial and set up your first revenue-goal split test, or read the revenue A/B testing overview first. If subscriptions are only one of your revenue lines, Humblytics for digital products covers the rest of the picture.
Sources
- beehiiv, The State of Paid Newsletters 2026 (June 22, 2026: $10/$100 medians unchanged since 2024, 0.62% median free-to-paid conversion, niche medians, annual billing crossing 50% of subscription revenue)
- beehiiv, Email A/B Testing Tool for Newsletters (native A/B testing covers subject lines only)
- MarketingExperiments, Price Testing (three-way $10 / $12.50 / $14.95 concurrent test, ~8,978 clicks per arm, $10 produced 32.8% more revenue)
- Conversion, Using experimentation to find a product's optimal price ($25 to $33/$41 monthly, net +16% revenue per visitor at 99% significance via cross-plan shift)
- Project C, Is it time to raise your subscription price? With The Charlotte Ledger ($9 to $12 monthly, 92% five-month retention, revenue +27%/+25%/+45%)
- Relijournal, Paid newsletter: the 200-subscriber math (€5/€8/€15 sequential prices, churn 9% vs 4.7%, LTV €55 vs €170)
- Ghost, How to price your subscription newsletter ($5-15/month band, Jessica Lessin quote, ImpactAlpha at $399/year)
Ship your first A/B test before this article ends.
Visual editor, no dev ticket, agent picks the winner. Dedication Agents lifted conversion 28% on their first test. ToForm went from 2% to 8% bookings.