"We ran a coupon and CVR went up, but revenue did not." "We raised the free-shipping threshold and AOV went up, but order count dropped." EC operators hear these almost every month. The cause is usually the same: CVR and AOV tend to move in opposite directions, and the team is only watching one of them.
This article focuses on lifting CVR and AOV at the same time, covering the trade-off structure, four compatible domains, phase-based priorities, and measurement pitfalls.
TL;DR
- CVR and AOV are two factors in
Revenue = Sessions × CVR × AOV. Pushing one usually drops the other - Compatible tactics fall into four domains that lift both at once: recommendation accuracy, value-bundle design, pre-purchase information, post-purchase follow
- Priorities shift by business phase: early-stage protects CVR, scale-up adds AOV, mature drives LTV
- Joint-axis judgment: use
RPS (Revenue Per Session) = CVR × AOV
1. The Revenue Decomposition
CVR is the share of visitors who buy. AOV is the per-order revenue. EC revenue decomposes into these plus session count.
Sessions sit on the acquisition side (ads, SEO, brand search). CVR and AOV are on-site experience metrics, and their tactical levers overlap. That overlap is why pushing one usually moves the other in the opposite direction.
2. Why Joint Lift Is Hard — Four Trade-offs
Four representative trade-offs where buyer psychology pulls in opposite directions:
| # | Tactic | CVR | AOV | Why it conflicts |
|---|---|---|---|---|
| 1 | Heavy discount coupons | up | down | Lower buy threshold but smaller per-order revenue |
| 2 | Raise free-ship threshold | down | up | Visitors who fall short of the threshold drop |
| 3 | Push high-price bundles | down | up | Single-item buyers drop |
| 4 | Push low-price items | up | down | Easier to convert but average order shrinks |
Baymard Institute research shows cart-abandonment reason #1 is "extra costs like shipping are too high" (48%). Most "lift one metric" tactics drop the other; joint lift needs a different category of tactics.
3. Four Domains That Lift Both
Tactics that lift CVR and AOV together share one shape: they do not block purchase intent and they add per-order value.
- A Recommendation accuracy: "Frequently bought together" modules on product / cart pages. High-precision cross-sell delivers AOV +10–30%
- B Value-bundle design: Not discount-bundles. Sell combinations single items cannot ("morning + night skincare set," "3-variety coffee tasting set"). Value, not price
- C Pre-purchase information: Stock status, delivery dates, return policy. Undisclosed shipping costs strongly correlate with cart abandonment
- D Post-purchase follow: Cart-recovery emails, post-purchase complement emails, member-only early access. Lifts second-and-onward AOV plus LTV (+20–40% in some studies)
4. Priority — Where to Start
Sequence the four domains by business phase. Running AOV tactics before traffic stabilizes drops CVR and shrinks revenue.
- Early-stage (sessions unstable): Protect CVR first. Start with Domain C (pre-purchase information)
- Scale-up (sessions stable): Add Domain A (recommendation) and Domain B (value-bundle). Add AOV without dropping CVR
- Mature (high repeat rate): Domain D (post-purchase follow) becomes central. Membership programs grow LTV
5. Measurement Pitfalls
Even with the right tactics, broken measurement makes the "up / down" call wrong.
| # | Pitfall | Fix |
|---|---|---|
| 1 | Bots in CVR denominator | Recalculate using bot-filtered sessions |
| 2 | AOV using pre-discount price | Net revenue (post-discount) / order count |
| 3 | Ignoring device / channel splits | Decompose by device, channel, landing page |
Pitfall #3 is the silent one. Mobile AOV typically runs 20–40% below desktop. A rising mobile share alone makes overall AOV look like it is dropping.
The cleanest joint view is RPS (Revenue Per Session) = CVR × AOV. If AOV rises but CVR drops by the same magnitude, RPS is flat and the tactic delivered no real joint lift.
Summary
- Pushing CVR or AOV in isolation tends to drop the other
- Joint-lift tactics fall into four domains: recommendation accuracy, value-bundle design, pre-purchase information, post-purchase follow
- Sequence by business phase: early-stage, scale-up, mature, with different priority domains in each
- Measurement: filter bots, use post-discount AOV, decompose by device / channel
- Use
RPS = CVR × AOVas the unified judge against measurement distortion
Detailed examples and the decision flow are in the original LP: How to Increase CVR and AOV Together.
Question for the community: When you found CVR moving in the opposite direction of AOV, what was the first tactic you tried to balance them? I'm curious whether Domain A (recommendation) or Domain C (pre-purchase info) is the more common first move in practice.