Ecommerce OS · Info Products7 min readUpdated November 2, 2026

Checkout A/B Testing for Info Product Businesses

Why course creators should A/B test the checkout, bumps and upsells before the sales page, the sample sizes involved, and tests not to run.

Nick Bonnassies, CMO at Tagada · checked against 8 sources

Article summary

TL;DR

Most course sellers don't have the traffic to A/B test a sales page properly. Start where conversion rates are high and every buyer counts: the order bump, the one-click upsell, the payment plan offer and the checkout form. Judge each test on revenue per visitor after refunds, and never test a fake discount, a resetting timer or a hidden renewal.

To detect a 20% lift on a sales page converting at 3%, you need about 13,900 visitors per variation; on an order bump taken by 30% of buyers, about 960 buyers (standard calculation, 95% confidence, 80% power).
SamCart reports that order bumps convert at "30-40% on average" on its platform, a vendor figure.
A "was" price in a test must be the "actual, bona fide price" you charged "for a reasonably substantial period of time" (16 CFR 233.1).
For memberships, ROSCA requires all material terms to be disclosed "before obtaining the consumer's billing information" in every variant.

Want to go further? Ask

On this page

This guide is part of our playbook on the best ecommerce platform for info product businesses. The general guide, for any business: custom ecommerce checkout.

Why test the checkout before the sales page

Sample size depends on the conversion rate you test. A sales page converts a small share of visitors, so small lifts take huge traffic to detect. A bump or an upsell is seen only by buyers, but a large share of them say yes.

What you test	Baseline rate (example)	Lift to detect	Needed per variation
Sales page, visitor to buyer	3%	+20% (to 3.6%)	about 13,900 visitors
Sales page, visitor to buyer	3%	+50% (to 4.5%)	about 2,500 visitors
Checkout, visitor to buyer	10%	+20% (to 12%)	about 3,800 checkout visitors
Order bump take rate	30%	+20% (to 36%)	about 960 buyers
Upsell take rate	30%	+10% (to 33%)	about 3,800 buyers

Calculated with the standard two-proportion formula, 95% confidence and 80% power. The baseline rates are illustrative, not benchmarks: use your own.

Fig. 1Thousands needed per variation, from sales page to bump

Sales page, +20%visitors, at 3% baseline13.9k
Sales page, +50%visitors, at 3% baseline2.5k
Checkout, +20%checkout visitors, at 10% baseline3.8k
Upsell, +10%buyers, at 30% baseline3.8k
Order bump, +20%buyers, at 30% baseline0.96k

Source: Two-proportion formula, 95% confidence, 80% power

The higher the base rate, the smaller the sample.

Calculated with the standard two-proportion formula, 95% confidence and 80% power. Baseline rates from the table above, illustrative.

So a creator selling 600 courses a month can finish a two-variant bump test in about three months (about 1,920 buyers), but would need far more traffic to test a sales page headline for a modest lift. Before that point, use qualitative research on the sales page: recordings, buyer surveys, support questions.

What to test, in order

#	Element	Variant to try	Metric
1	Order bump	Workbook vs templates pack; price; one bump vs two	Bump take rate, revenue per buyer
2	One-click upsell	Next course vs coaching call; order of offers; a downsell on decline	Upsell take rate, refunds at 30 days
3	Payment plan offer	Pay in full shown first vs plan shown first; 3 vs 4 payments	Revenue per visitor, installments collected
4	Checkout form	Fewer fields; Apple Pay and Google Pay on top	Checkout completion rate
5	Sales page	Headline; proof placement; price block	Visitor to checkout rate, only with enough traffic

How to design the bump and the upsell: order bumps and one-click upsells for info product businesses.

Example

A $297 course tests two bumps at $27: "Workbook" vs "Templates pack". With 1,000 buyers per variant, the templates bump is taken by 34% against 28%. That is $1.62 more revenue per buyer at the same traffic, which you then check against refunds.

Fig. 2Two $27 bumps on a $297 course

Workbook bump

Checkout$297 course
offersBumpWorkbook, $27

Taken by 28% of buyers

Templates pack bump

Checkout$297 course
offersBumpTemplates pack, $27Approved

Taken by 34% of buyers, $1.62 more per buyer

Same traffic, $1.62 more per buyer.

Illustrative example.

Testing price and payment plans

Price tests are the most valuable and the easiest to get wrong.

Test the presentation before the price
Pay-in-full first or plan first, total shown or monthly shown.
Test real prices only
Each variant must be a price you are willing to honor for every buyer who sees it.
Keep plan totals equal to pay-in-full
In every variant. Visa forbids "interest, or imputed interest" on merchant plans6. Plan design: payment plans for info product businesses.
Measure collected revenue
A plan variant that sells more but collects less is a loss.

Tests you should not run

A variant that wins by misleading buyers wins sales and loses them later as refunds, disputes or a processor review.

Variant	Why not
A "regular price" you never charged	A former price must be "the actual, bona fide price" offered "for a reasonably substantial period of time" (16 CFR 233.1)2
A timer that resets on reload	The FTC lists "countdown timers designed to make consumers believe they only have a limited time to purchase" as a dark pattern (FTC, September 2022)3
Membership terms moved below the button	ROSCA requires material terms disclosed "before obtaining the consumer's billing information" and "simple mechanisms" to stop charges (15 U.S.C. § 8403)4
An upsell with no visible "No thanks"	Stripe prohibits "high-pressure upselling" (Stripe)5
A testimonial that promises income	Stripe prohibits "deceptive testimonials" and "sites that promise fast and easy money" (Stripe)5

How to read a result

Fix the sample size before you start, and don't stop early because a variant looks ahead.
Split by buyer, not by session, so a returning visitor sees the same variant.
Count purchases from the server, deduplicated with the browser pixel, so each variant's sales are counted once7.
Compare revenue per visitor, not only conversion rate: a cheaper variant can convert more and earn less.
Check refunds and disputes 30 days later before you roll the winner out.

Decide the sample, the split and the metric before the test starts; check refunds before the rollout.

Where testing fits in the funnel: sales funnels for info product businesses and landing pages.

How Tagada helps

Run bump, upsell and checkout tests where the payment happens

A/B tests built into the checkout, on order bumps, one-click upsells and the checkout itself, in the same system that takes the payment.
Order bumps and one-click upsells "charged on the card just used, with no form", with "upsell on accept, downsell on decline".
Server-side events to Meta, Google and TikTok, so each variant's purchases count once.
Payments that keep test traffic converting: TagadaPay reports a "+18% approval uplift" from routing: "Route to the processor most likely to approve it, retry on decline"8. "If you sell something real, we will get you paid."
Pricing: "From 2.9% + $0.29", "No setup fees, no monthly minimums"8.

"Live in 24h. You own your vault." Apply today and go live in under 24 hours on TagadaPay; CRM plans are on the pricing page.

Apply today · Go live in under 24 hours

Start free

Run bump, upsell and checkout tests where the payment happens

Updated on November 2, 2026

Nick Bonnassies

CMO at Tagada

CMO at Tagada. Before Tagada, Nick built direct-to-consumer brands that generated over $115M in sales, raised over $25M in crowdfunding and sold two of them. He also works as a fractional CMO on marketing operations and AI systems.

Frequently asked questions

How much traffic do I need to A/B test a course sales page?

It depends on your conversion rate and the lift you want to detect. At 3% conversion, spotting a 20% lift takes about 13,900 visitors per variation with a standard calculation (95% confidence, 80% power). Below that, test the bump, upsell or checkout, where rates are higher and samples smaller.

What should I test first as a course creator?

The order bump, because buyers see it and a large share say yes, so tests finish fast. Then the one-click upsell, the payment plan presentation and the checkout form. Test the sales page headline once you have enough traffic.

Can I A/B test my course price?

Yes, if every price shown is real and honored for everyone who sees it. Start with presentation (pay-in-full first or plan first) before testing the amount. Don't test a fake "regular price", and measure collected revenue and refunds, not only sales.

Is A/B testing worth it for a small audience?

Testing the sales page usually is not until traffic grows. Testing the bump and upsell often is, since the base rates are high. For the rest, talk to buyers: surveys after purchase and questions to support show what to change.

Should I test one order bump or two?

You can test it. Measure revenue per buyer and checkout completion, not bump take rate alone: two bumps can raise order value but slow the checkout. Keep each bump clearly optional and priced.

Sources

Checked November 2, 2026

Tagada

Test the bump and the upsell before the sales page

Funnels, one-click upsells, subscription billing and payment routing, on one customer record.

Start free See pricing

01Create your free account
02Connect your store or build a checkout
03Launch funnels, upsells and subscriptions