conversion · popups · ab testing

Discount Popup vs Sticky Checkout Button: What to Test

Should you use a discount popup or a sticky checkout button? Learn when each fits, how to protect your margin, and how to test for orders instead of clicks.

By cartt.io Team · November 16, 2026 · 11 min read

Discount Popup vs Sticky Checkout Button: What to Test

You open your store on your phone, and an email offer covers the product before you've read its price. A discount popup can earn a subscription and encourage a purchase, but it can also interrupt a buyer and reduce the margin on an order you would've received anyway. Test one when you need to capture interest from visitors who aren't ready to buy. Test a sticky checkout button when buyers lose access to the next step while scrolling. Judge both by completed orders, revenue per visitor, and contribution after costs.

Do discount popups work?

A popup exchanges something you can deliver for an email address and, sometimes, useful information about intent. The address has potential value. The discount has an immediate cost when redeemed. Whether that trade pays depends on what happens after signup and what would've happened without the offer.

Imagine you're buying a replacement item you already know you want. A first-order offer appears on the way to checkout. You subscribe, use the code, and buy. The signup dashboard looks successful, yet the store may simply have charged you less for the same purchase.

That's why you need a comparison group that sees no popup. Comparing subscribers with everyone else won't answer the question: people who subscribe have selected themselves into that group. Include all eligible visitors assigned to each version, even those who never submit the form.

Also separate an existing subscriber from a new prospect. Suppress known subscribers and previous purchasers from an acquisition offer where your tools support it. Klaviyo supports audience targeting for recognized profiles, but its documentation explains that unidentified visitors can still see a form. Cleared browser data or a different device can leave a returning customer unrecognized. Suppression reduces repetition; it doesn't enforce first-order eligibility at checkout.

Early display and delayed display solve different problems. If your ad promises an offer on arrival, make that offer immediately accessible, preferably within the landing page. A mobile visitor following that promise shouldn't have to guess when it will appear. If you're selling a considered purchase and the visitor is still comparing materials or reading specifications, a delayed invitation or small teaser gives them room to evaluate it.

Neither your dislike of popups nor a high opt-in rate settles which approach earns more.

When should a popup show?

Start with the page and traffic segment you're testing. An average across your homepage, product pages, and checkout won't describe how long a new mobile visitor spends reading one product.

For a first email popup delay experiment, use that page's average time on page minus a few seconds as a candidate. This is a starting hypothesis, not a validated optimal-delay formula. An average hides short visits and long visits, and your analytics tool's definition matters. If the result would interrupt the opening sentence, try a teaser or interaction trigger instead.

Another candidate is the second page view. For visitors browsing products, you could test an exit intent popup after they've viewed at least two product pages. Keep these as separate experiments. Otherwise you won't know whether timing, audience, or the offer caused the difference.

Check what your form builder actually counts. Klaviyo supports time, scroll, page-count, and exit-intent triggers. Its page counter tracks unique pages within the browser tab's session; a generic page count doesn't establish that someone viewed product pages. Combine conditions deliberately and verify product-page tracking if that distinction matters. Requiring every condition is different from allowing any condition to trigger the form.

Exit intent is also an estimate. In Klaviyo, desktop detection uses cursor movement outside the visible page, while mobile detection uses a rapid upward scroll. A visitor scrolling back to a product photo might trigger it. Test on a real phone before treating it as a last chance to reach someone leaving.

A teaser leaves a small invitation available for the visitor to open. Klaviyo offers this, but a teaser doesn't necessarily disable automatic opening: the form can still appear when its timing conditions are met. Check both behaviors and keep the teaser clear of purchase controls.

Google Search Central recommends small banners that leave content accessible and warns that intrusive interstitials can hurt search performance. A common tip is that any delayed popup is safe for SEO. A timer doesn't establish that. Keep the product readable, provide an obvious close control, and avoid covering the entire page.

What can you offer besides a percent off?

Choose a benefit connected to what you're selling. Early access suits a real upcoming release. Drop reminders suit products customers wait for. A useful guide can help someone choose or use the product. These offers need a delivery plan: a vague VIP label gives someone little reason to stay subscribed.

Shopify's own email guidance includes early access and exclusive content as welcome incentives. Neither removes the work of creating useful content or organizing the release. Don't promise priority access if subscribers receive the same announcement at the same time as everyone else.

Free shipping is another option, provided you've priced the shipping cost. Shopify supports shipping discount codes and automatic shipping discounts, with conditions such as minimum spend and eligible countries. State those conditions beside the offer. Free shipping can cost more than the percentage discount it replaces, especially on an expensive delivery route.

Use this hypothetical example to compare offers. Suppose an item sells for $72 and has $36 in variable costs, including fulfillment and payment costs. A $9 discount leaves $27 contribution per order before advertising and overhead, compared with $36 at full price. Assume those costs stay fixed for this illustration.

Hypothetical resultNo popupDiscount popup
Eligible visitors1,2001,200
Completed orders2430
Revenue after discounts$1,728$1,890
Revenue per visitor$1.44$1.575
Contribution before advertising and overhead$864$810

Here, assume every order in the popup version uses the discount. More orders and higher revenue per visitor still produce less contribution. You'd need 32 discounted orders to match the original $864, before extra software costs or returns. Substitute your actual costs and redemption mix; this isn't a forecast.

A question can make signup more useful too. If you sell equipment, ask which activity the visitor needs it for, then use the answer to send relevant guidance. Shopify Forms supports custom choice fields. An extra field adds effort, so keep it only if you'll use the answer. A short form doesn't automatically become a recommendation quiz with branching results.

Finally, you can collect marketing opt-ins at checkout and skip the popup. Shopify provides a marketing consent option in checkout settings. This reaches visitors who've progressed further toward buying and leaves earlier browsing uninterrupted, but misses those who leave before checkout. Keep the marketing choice distinct from providing an address for an order.

For welcome-offer economics and the risk of teaching customers to wait for a deal, see fake discounts vs honest pricing. If you retain a code, review how to stop discount code leaks.

Does a sticky checkout button raise conversion?

You scroll through a long mobile product page, check the measurements, and read the returns explanation. You're ready to continue, but the purchase button is several screens away. Keeping that action accessible can remove the search for it. Whether it produces more completed purchases needs testing.

Placement determines what you're testing:

PlacementAction to keep accessibleWhat must still work
Product pageAdd the selected item to the cartVariant selection, quantity, and availability
CartProceed to checkout with the current cartItem edits and updated totals
Order-review pageContinue with the reviewed orderRequired choices, validation, and accurate charges

A sticky add to cart button and a sticky checkout button aren't interchangeable. The first changes the cart; the second advances a purchase already being assembled. Label each action accurately. A button that opens checkout shouldn't imply that payment has already happened.

Start where you can reproduce the problem. Scroll until the original action disappears and try to continue. If the page is short and the button is already visible, you may have little to gain. On a longer page, consider showing the extra bar only while the original control is out of view.

The bar must use the current selection and preserve validation. Test it after changing a variant or quantity, and with an unavailable item. Leave enough space around content, check the mobile keyboard, and tab through the page. W3C explicitly identifies sticky headers and footers that completely hide focused controls as an accessibility failure.

Don't budget for a dramatic conversion lift from this change. A before-and-after jump can reflect a repaired button, different traffic, a seasonal promotion, or another checkout change. A sticky action also can't answer an unclear delivery promise or make an unexpected charge acceptable. The broader mobile checkout optimization guide covers the surrounding form and layout work.

How do you test either one honestly?

Use the existing page as control and change one thing. Compare no popup with one popup, or the existing purchase control with an added sticky control. Testing both additions together can evaluate the package, but won't tell you which deserves credit.

Before you A/B test checkout changes, write a short plan:

  • Define the eligible visitors, random split, and consistent assignment for returning visitors.
  • Choose revenue per visitor as the primary measure, with orders and contribution as checks.
  • Set the sample requirement, minimum duration, purchase follow-up window, and latest end date.
  • Define the decision rule and the faults or losses that require an early shutdown.

Use the same revenue definition and follow-up window for both groups. Count revenue after discounts and account for refunds consistently. For an email offer, allow time for the promised follow-up to produce purchases. Klaviyo's form revenue reporting attributes purchases within a lookback window; that attribution doesn't establish that the popup created additional sales.

For sample planning, use your baseline purchase rate and the smallest improvement worth acting on. Optimizely's public calculator estimates visitors per variation for a conversion-rate test using its Stats Engine. It doesn't calculate the sample needed for your revenue-per-visitor metric. Revenue tests need a method that accounts for variation in order value.

The useful sample-size rule of thumb is that smaller effects need more evidence. There isn't a universal order count that makes a result reliable. If splitting traffic leaves each version with only a handful of purchases, a few orders can change the apparent winner. Estimate the traffic requirement before spending weeks testing a subtle change.

Optimizely recommends covering at least a full business cycle. Run across complete weeks for a store with a weekly pattern, and longer when the planned sample or buying cycle requires it. A week includes different shopping days; it doesn't guarantee enough evidence. Inspect mobile and desktop separately, but remember that splitting results also reduces the evidence available in each group.

Write the stop rule before seeing results. For a fixed-horizon test, finish the planned sample and duration before declaring a winner. A sequential testing system has different stopping rules designed for repeated evaluation. Follow the method your tool actually uses; checking an ordinary fixed-sample test daily and stopping on a favorable result isn't equivalent.

Keep ads, prices, and other page changes steady where possible. Record unavoidable changes and inspect whether an early gain fades as returning visitors get used to the new control. When traffic is too low, use before-and-after results alongside direct usability checks to make a cautious decision. You can fix an obvious obstruction without claiming you've measured a small conversion lift.

FAQ

Do popups hurt SEO?

Intrusive overlays can. Google advises keeping content accessible and using small banners instead of full-page interstitials. Neither a delay nor a later page view guarantees that your implementation is acceptable.

What delay should I use?

Use page-specific behavior to choose a first candidate, then test it. Average time on page minus a few seconds is a heuristic, not a proven optimum. Make an ad-promised offer accessible on arrival; give browsing visitors room to read.

Is a sticky button the same as sticky add to cart?

Sticky describes visibility while scrolling. Add to cart describes the action. A sticky control might add a product, open the cart, or proceed to checkout. Test the action and placement you actually need.

How long should an A/B test run?

Cover the relevant business cycle and meet your statistical method's evidence requirements, allowing for delayed purchases. Set an end date beforehand. If the result remains uncertain then, record it as inconclusive.

Do I need a discount popup at all?

No. You can use checkout opt-ins or an inline signup form, and spend your time fixing product information or purchase controls. Today, open your longest product page on your phone and complete a test purchase. Choose the first experiment from the obstacle you can actually find.