Testing a Free Shipping Progress Bar in the Cart

Progress bars toward free shipping are the most copied cart feature in ecommerce and the least honestly measured. How to set the threshold from your own data, run the test, and read the result without fooling yourself.

The threshold is the test, not the bar

A progress bar is a display. The economic decision is the threshold it counts toward, and most stores set that number by copying a competitor or picking a round figure near their average order value. Run your own distribution first: what share of orders land within ten dollars below your candidate threshold? That band is the only population the bar can move. If almost nobody sits just under the line, the bar is decoration, and if half your orders already clear it, you are not testing an incentive, you are announcing a shipping subsidy.

Set the threshold a realistic stretch above the median order, not the mean, which a few large orders drag upward. Then decide the success metric before launch: revenue per session is the honest one, because average order value will rise by construction while total profit quietly falls if the bar mostly rewards people who would have ordered anyway. A test whose metric cannot embarrass it is not a test.

Run it clean or do not run it

Split by session at the cart level, hold the control on the current cart, and keep the threshold, the copy, and the placement identical for the whole run except for the bar itself. The classic contamination here is seasonal: launching the bar the week a sale starts, then crediting the bar for the sale's traffic mix. If a promotion must run mid test, it runs for both arms or the test pauses. Anything else is storytelling.

Give it the time your traffic actually needs. A store doing a few hundred cart views a day needs weeks, not days, to separate a real lift from weekend noise, and the temptation to call it early peaks exactly when the first good weekend lands. Decide the sample in advance from your baseline conversion rate and the smallest lift worth having, then do not peek with intent. Peeking with intent is how a two percent wobble becomes a shipped feature.

What the bar changes besides order size

Watch the second order effects, because the bar moves more than the basket. Return rates on padded orders climb: the extra item added to clear the threshold is the first thing sent back, and it was often added with that plan. Track returns on bar arm orders that cleared the threshold narrowly. Watch also what gets added. If the stretch item is always your lowest margin product, the bar is a machine for selling your worst product at a shipping discount.

Support contacts are a signal too. Shoppers who miss the threshold by pennies write in, and the tone of those contacts tells you whether the bar reads as a game or a grievance. A bar that generates resentment at checkout is costing goodwill that no AOV chart shows. Read a sample of those tickets before you read the final results. Operators who skip the qualitative pass ship features their customers quietly hate with great statistical confidence.

Ship, raise, or kill it

Ship it if revenue per session rose and returns on narrow clear orders did not eat the gain. Raise the threshold if the lift is real but concentrated in orders that would have cleared a higher line; thresholds are a dial, and the first number you test is rarely the best one. Kill it if AOV rose while revenue per session sat still, because that pattern means the bar rearranged existing orders and paid shipping for the privilege.

Whatever you decide, document the threshold math and the result in the same note. The next person to touch the cart, possibly you in eleven months, should not have to reconstruct why the bar exists or rerun the test to defend it. Cart features accumulate like sediment in most stores. The lab habit of writing down the verdict is what keeps yours a cart instead of an attic.

Frequently asked questions

Where should a free shipping threshold be set?

A realistic stretch above your median order value, in the band where a meaningful share of orders land just below it. Set it from your own order distribution, not a competitor's number, and revisit it as your AOV moves.

How long should a progress bar test run?

Until it reaches a sample decided in advance from your baseline conversion rate and traffic. For most smaller stores that means weeks. Calling it after one strong weekend is the most common way these tests produce false wins.

What metric decides whether the bar worked?

Revenue per session, checked against return rates on orders that narrowly cleared the threshold. Average order value alone will mislead you, because the bar raises it mechanically even when it adds no new profit.

Can a progress bar hurt a store?

Yes. Set the threshold where most orders already clear it and the bar just gives away shipping. It can also inflate returns, since the item added to clear the line is the item most likely to come back.