I often get asked how to raise average order value (AOV) without tanking conversion rates. It’s a tension every marketer feels: you want customers to spend more, but you don’t want to push them away with sticker shock. Over the years I’ve run dozens of pricing experiments and A/B tests that balance those goals — and I’ve learned that the right methodology is as important as the pricing idea itself. Below I walk you through a pragmatic, step-by-step approach to designing pricing A/B tests that increase AOV while protecting conversion.

Start with a clear hypothesis and measurable goals

Every successful experiment begins with a crisp hypothesis. Instead of “raise prices,” I frame it like this: “If we introduce a $X premium bundle with added perceived value, then AOV will increase by Y% while conversion rate remains within Z% of the control.” That gives you three measurable variables: the treatment (new price or offer), the desired change in AOV, and the allowed impact on conversion.

Define primary and secondary metrics up front. My go-to list:

  • Primary: AOV (average order value)
  • Secondary: Conversion rate, revenue per visitor (RPV), unit sales, and profit margin
  • Guardrail metrics: cart abandonment, bounce rate on pricing pages, and customer lifetime value (if you can model it)

    • Pick the right type of pricing test

    Not all pricing experiments are just “higher vs lower price.” Here are formats that often raise AOV without a conversion drop:

  • Anchoring tests — show a higher “original” price alongside a discounted or promoted option to create perceived value.
  • Bundling and tiering — introduce a premium bundle or subscription tier that increases AOV while giving users choice.
  • Decoy pricing — include a middle option designed to make the higher-priced option look like better value.
  • Shipping and threshold experiments — free shipping thresholds (e.g., free shipping over $75) can nudge AOV up.
  • Order bump or add-on offers — small, high-margin add-ons at checkout.

    • Each type affects user psychology differently. Bundles and order bumps often increase AOV with minimal impact on conversion because they add perceived value. Pure price increases are riskier.

    Segment your audience intelligently

    One-size-fits-all rarely works for pricing. I always segment tests by behavior and intent:

  • New vs returning customers
  • High-intent traffic (PPC, email clicks) vs. exploratory traffic (organic)
  • Cart value segments (people already near your free-shipping threshold)
  • Device (mobile vs desktop) — mobile users can be more price-sensitive

    • Often you'll find a tactic that works for returning customers but not for first-timers, so you can target it strategically rather than apply it site-wide.

    Calculate sample size and duration

    Underpowered tests are misleading. Use a sample size calculator with baseline conversion and the minimum detectable effect (MDE) on conversion and AOV. If your baseline conversion is low (e.g., 1-2%), you’ll need a large sample to detect meaningful differences.

    Practical rules I use:

  • Don’t run a test for less than 2 full business cycles (typically 2 weeks) to cover weekday/weekend effects.
  • Ensure you reach the calculated sample size for both conversion and AOV tests — sometimes AOV needs more users because it’s noisier.

    • Design the test and variants

    Keep changes limited to the price or price presentation to isolate effects. Common treatments I’ve used successfully:

  • Variant A: Control (current pricing)
  • Variant B: New bundle at +20% price but with 30% more perceived value (bonus product, priority shipping)
  • Variant C: Anchored option — show original higher price struck through with current price next to it

    • Use consistent copy, visuals, and checkout flow across variants. The only difference should be the price or the way it’s positioned.

    Choose the right statistical approach

    I prefer Bayesian methods for pricing experiments because they let you understand probability of uplift directly (e.g., "There's a 92% chance variant B increases AOV"). That said, classical frequentist tests work too if you stick to pre-specified significance levels and avoid peeking.

    Key practices:

  • Pre-register your primary metric and minimum detectable effect.
  • Don’t stop the test early because results look promising — this inflates false positives.
  • Consider sequential or adaptive methods if you need faster decisions, but use proper corrections.

    • Track the right KPIs with a QA plan

    Make sure analytics are solid before you launch. I always do an event-level QA checklist:

  • Price displayed matches price passed to cart and to backend orders
  • Conversion events fire once and at the right step
  • Revenue and tax/shipping calculations are consistent
  • Ensure AOV is computed the same way across variants (order-level vs item-level)

    • Run a smoke test with a small percentage of traffic first to catch integration and tracking issues.

    Analyze results beyond averages

    AOV is useful but noisy. I slice results by segment, device, traffic source, and product category. Some deeper analyses I run:

  • Distribution of order values — did the treatment shift a small number of huge orders or broadly lift basket size?
  • Impact on units per order — bundling may increase items per purchase even if conversion dips slightly.
  • Profitability — increased AOV is great, but margin matters. Calculate gross margin per order.
  • Long-term effects — for subscription or repeat-buy categories, model expected CLTV impact from changes.
    • Metric Control Treatment Lift
    AOV $45.60 $52.20 +14.5%
    Conversion rate 2.10% 2.03% -3.3%
    Revenue per visitor $0.9576 $1.0609 +10.8%
    Gross margin per order $18.24 $20.88 +14.5%

    Be ready to iterate and phase rollout

    Even a winning test sometimes needs tweaks. I usually do a staged rollout:

  • Start with a small percentage of traffic once the experiment wins statistically.
  • Monitor for week-to-week changes and for operational impacts (e.g., support tickets, returns).
  • Adjust messaging, bundle contents, or thresholds based on feedback and additional tests.

    • Watch for common pitfalls

    Here are traps I’ve fallen into and learned from:

  • Confounding changes — changing copy, visuals, and price simultaneously makes results ambiguous.
  • Ignoring segmentation — a variant that hurts first-time buyers but helps returning customers can be misunderstood if you only look at overall metrics.
  • Short test durations — seasonality or day-of-week effects can mislead you if the test is too short.
  • Not accounting for shipping/taxes — showing a higher price but hiding steep shipping until checkout will kill conversion.

    • Practical examples that worked

    I once tested a $10 “priority pack” add-on for an electronics accessory store. The add-on added only 10% to the AOV on average, but conversion didn’t drop because customers perceived the add-on as convenience. In another test I introduced a $25 bundle (two items + extended warranty) that increased AOV by 18% and slightly lowered conversion by 2%, but net revenue and margin per visitor both improved — making it a clear win.

    Operational considerations

    Before you flip a winner live, coordinate with fulfillment, customer service and finance. A higher AOV with a complex bundle can increase returns, change shipping profiles, or affect inventory. I always run a 1-2 week operational pilot after statistical wins to catch anything the analytics don’t show.

    Pricing experimentation is as much about psychology and positioning as it is about numbers. By building clear hypotheses, segmenting thoughtfully, ensuring robust sample sizes and analytics, and staging rollouts, you can lift AOV without harming conversion — and often improve overall profitability. If you’d like, I can help sketch a test plan tailored to your product mix and traffic patterns.