Price testing means running controlled experiments to find the price that maximises profit per visitor, not just conversions. Your next step is to pick one high-impact product and run a randomised A/B price test measuring profit per visitor. A field experiment with an online retailer showed what this can deliver, and the ACCC sets the display rules you must follow.
TL;DR:
- A retailer’s randomized field test lifted expected revenue by approximately 9.7% on tested items while still meeting its margin targets.
- Score candidates by traffic, revenue share, gross margin, inventory stability, and price sensitivity; exclude seasonal items, clearance stock, and products facing supplier price changes.
- Bandits can shift traffic toward the better price during short promotions, but uneven exposure makes them weaker for estimating lasting price elasticity.
- Set sample size and end date before launch, keep each visitor’s price consistent across pages and emails, and include variable costs in profit per visitor.
- In Australia, show one total price that includes unavoidable costs, substantiate genuine prior prices in discount claims, and keep records of what each shopper saw.
Table of Contents
- What is price testing and why does it matter for your store?
- Which products and SKUs should you test first?
- How do the main price testing methods compare?
- How do you run an A/B price test step by step?
- Which metrics should you measure, and how do you calculate them?
- How do you analyse results and pick a winner?
- What mistakes should you avoid, and which Australian pricing rules apply?
- What is your quick checklist for a first price test?
- What do senior ecommerce strategists focus on when testing prices?
- How can we help you test prices and build a pricing strategy?
- FAQ
- Sources
What is price testing and why does it matter for your store?
Price testing is a structured experiment: you show different prices to comparable groups of shoppers at the same time, then compare what each group did. That is different from an ad-hoc price change, where you move a price on Monday, see sales shift by Friday and credit (or blame) the price when a promotion, a holiday or a stock-out may have done the work.
The evidence that disciplined testing pays is solid. In a field experiment with an online retailer, randomised price variations lifted expected revenue by approximately 9.7% on the tested items while the retailer still met its margin targets. Harvard Business Review’s write-up of lessons from more than 1,000 ecommerce pricing tests reaches the same conclusion: controlled experiments produce measurable revenue gains, and randomisation is what lets you estimate price elasticity without bias.
Where you test also matters. A meta-analysis of 2,732 A/B tests found that price-promotion experiments and experiments on category or listing pages sit among those with the largest effect sizes, and that price experiments tend to bite harder earlier in the conversion funnel. In plain terms, shoppers compare prices while browsing, long before they reach checkout. A price shown on a listing page shapes who even clicks through, so testing there can move results more than tweaking a checkout banner.
The payoff for a store owner is practical. You stop guessing, you stop copying competitors, and you get a defensible number for what your customers will actually pay. For a wider view of how this fits your overall approach, see our explainer on how ecommerce pricing strategy works in 2026.
Which products and SKUs should you test first?
Start where a small percentage change moves real dollars. Testing a slow seller tells you very little, because it takes months to gather enough visitors. Field experiments in this area focus on high-impact items rather than the long tail, because that is where both the data and the money are.
Rank your candidates with a simple score. You do not need a model, just a spreadsheet with these columns:
- Traffic: unique visitors to the product page over a recent period. More visitors means a faster, cleaner result.
- Revenue share: the percentage of total store revenue the SKU generates. Top sellers and category leaders matter most.
- Gross margin: a healthy margin gives you room to test both higher and lower prices without risking a loss.
- Inventory stability: pick products you can keep in stock for the whole test, so a stock-out does not corrupt the result.
- Price sensitivity signals: frequent discount requests, cart abandonment on that SKU or heavy competitor comparison suggest a price that might be off.
Give each SKU a score from 1 to 5 on each column, add them up and test the top one first. Leave out anything seasonal that is about to peak or fade, anything on clearance and anything with a supplier price change coming.
One more filter: avoid products shoppers routinely compare as a bundle with others in your range. Their prices interact, and you want a clean read on one change at a time for your first test.
How do the main price testing methods compare?
There is no single best method. Each one trades certainty against speed, and the right pick depends on your traffic and what you need to learn.
A/B price testing splits visitors randomly between two or more prices. It gives the clearest causal answer and works best when you have a measurable goal such as profit per visitor. The cost is patience: you wait for enough data, and some visitors see the weaker price.
Multi-armed bandits shift traffic toward the better-performing price while the test runs. You capture uplift faster, which suits short promotions. The catch is that uneven traffic can bias long-term elasticity estimates, so bandits are weaker when you want a lasting pricing rule.
Price tiers and anchoring tests change how a price looks next to alternatives, for example showing a premium option beside your core product. They can shift perceived value without a discount.
Bundles and shipping tests are worth separating from product price tests. Shoppers react to the total they pay, so a lower list price with higher shipping may behave differently to the reverse. Test one lever at a time.
Dynamic rules and competitor scraping adjust prices automatically based on inventory, time or rivals’ prices. They are tempting, but research shows that competitor-matching heuristics miss product-specific demand and should account for inventory and cross-price elasticity. Treat any automated rule as a hypothesis and validate it with a controlled test before it touches your whole catalogue. The ACCC confirms dynamic pricing is legal in Australia but expects clarity about total cost and the ability to substantiate price increases.
A quick way to choose:
- Need a durable answer for a core product? Run a classic A/B test.
- Running a short campaign with limited time? Consider a bandit and accept rougher estimates.
- Worried about discounting eroding your brand? Test tiers or bundles before cutting prices.
- Tempted by automation? Validate the rule in a controlled test first.
If you are also testing product listing copy alongside price, this partner guide on ecommerce keyword research helps you keep traffic sources steady between test arms.
How do you run an A/B price test step by step?
Follow this workflow from hypothesis to rollout.
- Write a falsifiable hypothesis and pick one KPI. For example: “Raising the price of SKU X will not reduce profit per visitor.” Profit per visitor beats conversion rate because a higher conversion at a lower price can still lose money.
- Set a price floor. Work out your cost per unit including product, shipping, payment fees and expected returns, then add the minimum margin you will accept. No test arm goes below that floor.
- Choose your arms. Keep it simple: your current price as the control and a small number of variants close to it. Wide jumps tell you little and risk shopper confusion.
- Randomise cleanly. Assign each visitor to an arm using a cookie or a server-side flag, and make sure the same person always sees the same price across sessions and devices where you can identify them.
- Keep the price consistent end to end. The product page, cart, checkout, emails and any ads must show the same price for that visitor. A mismatch breaks trust and breaks the data.
- Estimate sample size and minimum duration. Use a sample size calculator with your baseline conversion rate and the smallest difference you care about. Run long enough for weekday and weekend behaviour to appear, and do not stop the moment one arm pulls ahead.
- Run, monitor and log. Watch for stock-outs, competitor sales, ad budget changes or email sends that hit one arm harder. Record every such event with a date so you can explain odd results later.
- Analyse once, at the planned end. Compare profit per visitor between arms, then follow the analysis steps below before you roll anything out.
Pro Tip: Fix your sample size and end date before launch and write them down. Checking daily and stopping at the first good-looking number is the most common way small stores fool themselves.
Small stores often worry they lack the traffic. If a SKU cannot reach a decent sample in a reasonable period, test a bigger price gap, pool similar products or move up to a category-level test rather than shortening the run.
Which metrics should you measure, and how do you calculate them?
Optimise for profit, not vanity numbers. These are the core metrics and their simple formulas:
- Conversion rate: orders divided by visitors. Useful, but blind to price.
- Revenue per visitor (RPV): total revenue divided by visitors. It captures both conversion and order value. Our guide to conversion benchmarks and RPV shows how to read it against your category.
- Gross margin: (price minus cost of goods) divided by price.
- Profit per visitor (PPV): (revenue minus cost of goods, shipping, payment fees, platform fees and refunds) divided by visitors. This is the number that decides the winner.
Fold every variable cost into PPV. If a higher price attracts more returns, or a free shipping threshold changes, the arm that looked better on RPV may lose on PPV. Our overview of ecommerce performance metrics covers how to pull these figures from your analytics.
You can also estimate price elasticity from your test arms: divide the percentage change in units sold by the percentage change in price. A result below minus 1 means demand is elastic, so a price rise cuts volume by proportionally more and usually hurts revenue. A result between 0 and minus 1 means demand is inelastic, so a rise often adds profit. Elasticity matters beyond one SKU, since it tells you which parts of the catalogue can carry higher prices and which need sharper ones.
Randomisation is what makes that estimate trustworthy: randomised experiments are needed to estimate unbiased price elasticities, which is why comparing last month’s sales to this month’s is not a substitute for a real test.
How do you analyse results and pick a winner?
Wait for your planned end date. Looking early and stopping when a result looks good inflates false positives, because random swings are largest when data is thin.
When the test ends, compare PPV between arms and check the confidence interval, not just the headline difference. Check it before you call a winner, and be cautious if it includes zero. If the interval is wide and straddles zero, the honest answer is “no clear winner”, which is still a useful result: it tells you the price is not your lever, and you can move on.
Next, look for segment effects. An overall flat result can hide a win in one group and a loss in another. Break results down by:
- traffic channel (paid social, search, email, direct)
- new versus returning customers
- device type
- product attributes such as colour, size or brand
If a variant wins for returning customers but loses for new ones, you may be better off testing a different offer for each group than rolling out one price.
Rolling a single-SKU win out to the catalogue needs care. A price that works for your best seller may not suit related products, because items compete with each other and cross-price elasticity means changing one price shifts demand for its neighbours. Roll out in stages, apply the rule to similar products first and monitor PPV after rollout. Our guide on how to scale ecommerce profitably covers turning a test win into policy.
What mistakes should you avoid, and which Australian pricing rules apply?
Most failed tests fail for operational reasons, not statistical ones. Watch for these:
- Inconsistent price displays: a different price in the cart or in an email invalidates the result and can breach consumer law.
- Traffic leakage: shoppers who clear cookies, switch devices or share links see different prices, so arms blur together.
- Seasonal bias: running one arm during a promotion and another outside it compares unlike periods.
- Stopping too early: an early lead often evaporates as data accumulates.
- Chasing conversion at the expense of margin: a cheaper price nearly always converts better, which is why PPV is the safer target.
- Ignoring cross-price effects: lifting one product’s price can push demand to a cheaper sibling.
On the regulatory side, Australian Consumer Law applies to every arm of your test. The ACCC states that prices must be shown as a single total inclusive of unavoidable costs, and that “was/now” or RRP savings claims must reflect genuine prior prices. Its sweep of Black Friday advertising flagged questionable “was/now” pricing and site-wide claims with exclusions, and warned against inflating pre-sale prices to exaggerate discounts. The regulator’s guidance on false or misleading claims adds that sale prices that were never genuine and hidden fees both count as misleading. Keep a record of what each visitor was shown, and never use a test price as a fake “before” price later.
What is your quick checklist for a first price test?
Copy this list and tick it off before you launch.
- Write your hypothesis and choose profit per visitor as the KPI.
- Set a price floor that covers all costs plus your minimum margin.
- Pick one SKU using your traffic, revenue share and margin score.
- Estimate sample size and set a minimum duration that allows weekday and weekend behaviour to appear.
- Confirm the price is identical across product page, cart, checkout and emails for each visitor.
- Check your ACCC compliance: total price shown, no misleading “was” prices.
- Reserve enough stock to cover the whole test.
- Schedule monitoring windows and a log for events such as promotions or stock-outs.
- Write the analysis plan: metrics, confidence level and segment cuts.
- Plan a staged rollout and monitor performance after rollout.
What do senior ecommerce strategists focus on when testing prices?
My view is that most stores test the wrong thing. They obsess over the headline conversion rate and treat a winning percentage as a win, when the number that pays the bills is profit per visitor. A price that improves conversion but drags margin down can be a loss with a nice chart.
The behaviour I trust is boring. Fix guardrails first: a price floor, a maximum change per test and a clear end date. Test by cohort, because new and returning customers rarely respond the same way. And accept null results, since “customers do not care about this price” frees budget for fixing something that does matter.
A controlled test can show whether demand changes when a premium product’s price rises. One lesson is that loud feedback is not the same as measured behaviour.
Run tests in-house when you have steady traffic and someone who can own the analysis. Bring in expert help when stakes are high, traffic is thin or results keep contradicting each other.
— Liza
How can we help you test prices and build a pricing strategy?
If you would rather not build this from scratch, we can run it alongside you. Our senior strategists work hands-on with your store, with no outsourcing, and every recommendation rests on your own data. Here is where we typically help with pricing work:
- Ecommerce strategy: deciding which products and segments to test, and how pricing fits your growth plan
- Conversion rate optimisation: setting up clean experiments and consistent pricing across your site
- Analytics and reporting: tracking profit per visitor, not just sales
- Workshops: a focused session to build your first test plan with your team
[image:cta_image]
Pricing and experiment support sits inside our ecommerce marketing service, and a workshop or audit of your top products is a sensible first step. Get in touch and tell us which SKU you want to test first.
FAQ
What is price testing and how does it work?
Price testing is a controlled experiment where different groups of shoppers see different prices for the same product. You compare results such as profit per visitor to find the price that earns the most. Randomisation is what makes the comparison reliable, as lessons from over 1,000 ecommerce pricing tests show.
What is the 80/20 rule in ecommerce?
The idea is that a small share of your products or customers may produce a large share of your revenue. For price testing, it means starting with your top sellers, because that is where a small price change moves the most profit and gathers data fastest.
How long should an ecommerce price test run?
Run it long enough for weekday and weekend behaviour to appear, and longer if traffic is low. Set the end date and sample size before launch, and avoid stopping early when one arm looks ahead.
Is it legal to show different prices to different shoppers in Australia?
Dynamic pricing is legal in Australia, but you must not mislead shoppers about the total cost or the reason prices change. Each visitor must see a clear, accurate total price. Check the ACCC guidance or seek legal advice for your specific setup.
Sources
- Analytics for an Online Retailer: Demand
- Price displays | ACCC
- Working Paper
- Lessons from More Than 1000 E-Commerce Pricing Tests
- An Empirical Meta-analysis of E-commerce A/B Testing Strategies
