By 
Most advice on Shopify price testing assumes you already have the traffic to run a clean live test. Most Shopify stores do not, which is why so many of them either never test price at all or run a test that never reaches an answer and read the noise as a result.
So here is the version that starts from where you actually are. There are two honest roads to a better price, one for stores with the traffic and one for stores without it, and the whole game is knowing which road you are on before you spend a penny. This guide covers both: how a live price test works and what it really costs in traffic, how to get a directional read on price when you cannot run one yet, and the sequence that wastes the least money getting from a guess to a number you trust.
What price testing on Shopify actually is
Shopify price testing means putting different prices in front of different shoppers, or different would-be shoppers, and measuring what that does to demand. There is a catch worth knowing before you start, though: the platform barely does it for you. Shopify’s own Smart Pricing app can now run native A/B price experiments, but it is in early access, US-only, needs a Grow plan or higher, and only unlocks for stores already doing real volume, ten or more products each selling twenty-five-plus units a month, and it will not touch discounts, subscriptions, multi-currency or more than one test at once. For almost everyone, then, a real live price test still runs through a third-party app that swaps the price at runtime, most of them built on Shopify Functions, which is why searching for a “shopify price test app” turns up a shelf of them rather than a setting in your admin.
That leaves you with two roads, and they answer different questions.
The first is live testing. You have the traffic, you split it, you show real buyers real prices, and you watch which one makes more money. This is the strongest evidence there is, because nobody is guessing about what they would do. They are doing it.
The second is pre-spend testing. You do not have the traffic yet, or you do not yet know the rough range, so you measure willingness to pay before you commit, narrow the field to a couple of sensible prices, and only then put them live.
Most guides pick a side. The right answer depends entirely on one number.
The one number that decides your road
That number is how much traffic, and therefore how many orders, you actually do.
A live price test only tells you something true once each price has collected enough conversions to be statistically significant. Below that line you are not reading a result, you are reading noise, and noise is happy to tell you to raise your price the week before revenue drops.
How much is enough? Start from a realistic Shopify conversion rate. The Littledata benchmark puts the typical store in low single digits, in the region of 2 to 3 per cent. Feed that into Evan Miller’s sample-size calculator, the tool most practitioners use, and the maths is sobering. To detect a normal price move, something that shifts conversion by about 10 per cent, you need roughly 50,000 visitors per price, so around 100,000 visitors in total for a simple two-way test. At a 3 per cent conversion rate that is about 3,000 orders spread across the test. A big, obvious price gap gets there far faster. A subtle one can need four times the traffic.
There is a trap inside the waiting, too. Do not stop the test the moment it flashes significant. Peeking and calling a test early is one of the most reliable ways to fool yourself, because with enough looks almost any test crosses the line by chance. Pick your sample size in advance and let it run.
So the decision is not really about tools. It is about whether you can reach that number in a sensible window.
| Where you are | Your realistic move |
|---|---|
| Comfortably over ~3,000 orders a month | Run a live price test. You will get clean reads in weeks. |
| A few hundred to a couple of thousand orders a month | Test only big, obvious price gaps live, or get a pre-spend read first and test rarely. |
| Under a few hundred orders a month, or pre-launch | Do not run a live price test yet. Measure willingness to pay, act on the directional read, and revisit live testing as you grow. |
Rough figures, not gospel. Your product, your margin and how big a change you are trying to detect all move them. But the shape holds, and traffic is the gate.
Road one: price testing with the traffic
If you clear the volume bar, live testing is where you want to be.
Set it up through one of the price testing apps. Intelligems is the most complete for price, shipping and margin work; ABConvert and Elevate compete mainly on being cheaper; and Shoplift leans towards theme and content testing. Confirm current pricing on each vendor’s page, because it moves, and it usually scales with your order volume. If you want the full teardown of the category leader, our Intelligems review is the companion piece to this one.
Then measure the right thing. This is where most price tests go wrong: they watch conversion rate. A lower price almost always converts better, so conversion rate on its own will talk you into leaving money on the table every time. Judge a price test on revenue per visitor, and better still profit per visitor once you have loaded your cost of goods, because the only price worth shipping is the one that makes you the most money per person who lands, not the one that makes the most people buy. It is common for a test to show a lower price winning on conversion and losing on profit at the same time, which is exactly why the serious price-testing tools report profit per visitor as their headline number rather than conversion at all.
A few guardrails separate a real test from an expensive mess. Change one thing: if you move price and the free-shipping threshold at once, you will not know which did the work. Give it a fixed run: decide the sample size and the end date up front, and hold your nerve. And watch downstream, because price flows through cart, checkout, ads and any product feeds, so a price test can quietly move your return on ad spend and your margin, not just the number on the product page.
Road two: price testing without the traffic
If you cannot reach significance in a sensible window, live testing is the wrong first move. You would spend a quarter of live revenue narrowing a range you could have mapped in days. Do this instead.
Measure willingness to pay before you commit. There are two well-worn survey methods for it. The Van Westendorp Price Sensitivity Meter, introduced by the Dutch economist Peter van Westendorp in 1976, asks people at what price a product feels too cheap, a bargain, expensive but still worth it, and too expensive, and reads an acceptable band out of where those answers cross. The Gabor-Granger method does the more direct thing: it asks whether someone would buy at a specific price, then moves the price up or down, and builds a demand curve you can read a revenue-maximising point off. Van Westendorp maps the range; Gabor-Granger sharpens the point inside it.
Both share the same honest limit. They ask people what they would do, and stated intent is softer than a live purchase. Treat the output as a directional read that tells you which prices are worth testing, not a guaranteed sales number. That is exactly what it is for.
This is also where a pre-spend purchase-intent read earns its place. Rather than survey the open market, you put the decision in front of an audience built to look like your buyers and see how intent moves as the price moves. With TestFeed you can put an in-context price in front of an audience like your buyers and get back a purchase-intent read, the reasons in the shoppers’ own words, and a clear next move, in days rather than weeks. It is a pre-spend, directional signal, not a sales forecast and not a substitute for a live test once you have the volume to run one. Use it to pick the two or three prices worth putting live, then let live traffic settle the winner.
The sequence that wastes the least money
The mistake I see most is treating one road as the whole map. Discovery and proof are different jobs, and the cheapest path uses them in order.
First, discover the range. If you are staring at a new product with no idea whether it is a 25 item or a 60 item, do not burn live weeks finding out. Map willingness to pay and get to a sensible band fast.
Second, shortlist. Turn that band into two or three real candidate prices, not ten. Every extra price you test multiplies the traffic you need.
Third, prove it. Once you have the volume, run those two or three live and let revenue per visitor pick the winner.
Here is what that looks like in practice. Say you are launching a supplement and your gut says somewhere between 30 and 50. Guessing, you might test 34 against 39 live, wait a quarter for a wobbly read, and never learn that plenty of buyers would have paid 45. A pre-spend read on the whole 30 to 50 band points you at 39, 44 and 49 as the interesting prices and rules out the bottom of the range as money left on the table. Now your live test is three sensible candidates instead of two random ones, and you already knew the floor was too low before you spent a penny proving it. Same tools, far less waste, because you discovered before you tried to prove.
The guardrails: fairness and the law
Showing different shoppers different prices makes people nervous, so be clear on where the line sits.
Randomised price experimentation, where the price is shown plainly before anyone buys and the difference is not based on who the person is, is standard and legitimate practice. It is not the same thing as individualised pricing set from someone’s personal data, browsing or location, which is what regulators have started scrutinising. The FTC’s 2025 surveillance pricing study is about that second thing, not about a clean A/B test. The distinction is real, but a customer who finds out a friend paid less will not care about it. So keep test windows short, honour the lower price if someone asks, and think twice before running price tests on loyal repeat buyers who will notice and remember.
Where to start
If you do the volume, pick a price testing app, test one thing at a time, judge it on profit per visitor, and hold the test to its planned sample size. If you do not, measure willingness to pay first, take the two or three best prices forward, and only run live once the traffic can carry a clean read. Either way the winning move is the same: know which road you are on before you spend, and never let a low price flatter you on conversion while it quietly costs you money.
Frequently asked questions
Can you A/B test prices on Shopify?
Yes, but not with a native Shopify feature. Shopify will not show different prices to different visitors on its own, so you run the split through a price testing app such as Intelligems, ABConvert or Elevate, most of which use Shopify Functions to swap the price at runtime. If you do not have the traffic for a live test, you measure willingness to pay off-site instead.
How much traffic do you need to price test on Shopify?
Enough conversions per price to reach statistical significance. At a typical 2 to 3 per cent conversion rate, detecting a normal price move needs roughly 100,000 visitors in total for a two-way test, around 3,000 orders. A big, obvious price gap needs far less; a subtle one far more. Below that volume, get a pre-spend read before you test live.
What is the best Shopify price test app?
It depends on your volume and budget. Intelligems is the most complete for price, shipping and margin testing at scale; ABConvert and Elevate compete on price; Shoplift is stronger on theme and content than on price. Confirm current pricing on each vendor’s page, since all of them scale with order volume.
Is it legal to show different customers different prices?
Randomised price tests, where the price is transparent before purchase and not set from a person’s personal data, are standard and legal in most markets. That is different from individualised, data-driven pricing, which regulators are scrutinising. Keep windows short, honour the lower price if asked, and avoid testing prices on loyal repeat buyers.
How do you test price without traffic?
Measure willingness to pay before you commit. Use a pricing survey method like Van Westendorp or Gabor-Granger, or a pre-spend purchase-intent read against an audience like your buyers, to find the two or three prices worth testing. Then run those live once you have the volume to reach a clean result.
