The average order value playbook
AOV is the growth lever that needs no new traffic. The failure mode is copying tactics — bundles, thresholds, upsells — without the data that decides whether they fit your catalogue.
Every AOV tactic works somewhere and fails somewhere else. Free-shipping thresholds work when your order distribution clusters below a natural jump; bundles work when co-purchase affinity is real; checkout add-ons work for low-consideration accessories. This playbook derives which levers your data supports before building any of them — then builds the supported ones properly.
Stage 1: Read the order distribution
Pull the distribution of order values and the co-purchase matrix. Three facts decide the plan: where orders cluster relative to candidate shipping thresholds, which products are genuinely bought together, and what share of orders are single-item where an obvious companion exists. That last number is usually the embarrassment — and the opportunity.
Order value histogram and its natural clusters
Co-purchase affinity per product and category, from real orders
Single-item order share where a companion product exists
Margin per candidate add-on — AOV that costs margin is theatre
Stage 2: Build the supported levers
A shipping threshold sits just above the largest order cluster, announced where it changes behaviour: cart and product page, with a progress indicator. Bundles come from actual affinity pairs, priced so the discount trades against the second item’s margin, not the first’s. Add-on placements go where consideration is already resolved — cart and post-add moments — never interrupting the primary decision.
Stage 3: Extend past the checkout
The highest-margin AOV move is often the second order, not a bigger first one. Post-purchase sequences offering the companion at the point of ownership — the case a week after the phone, the refill before it runs out — raise cumulative order value without checkout friction. Measure the whole programme on margin per order and repeat rate, not AOV alone.
How it works
Derive the levers
Distribution, affinity, and margin analysis decide which tactics your catalogue supports.
Build them properly
Thresholds placed against real clusters, bundles from real pairs, add-ons where decisions are already made.
Measure margin, not vanity
AOV, margin per order, and repeat rate tracked together against the pre-change baseline.
What you get
Tactics chosen by your data rather than by blog consensus
A shipping threshold that moves the actual order cluster
Bundles with real affinity behind them, priced to protect margin
Post-purchase companions that raise lifetime value without checkout friction
Handled by these agents
Works with
Frequently asked questions
What is a good AOV?
Relative to your own baseline and margin structure — a higher AOV built on discounts can be a worse business. Track AOV, margin per order, and repeat rate as one panel.
Do free-shipping thresholds always work?
They work when a meaningful order cluster sits just below a reachable threshold. If your orders already clear it, or the gap is too wide to bridge with one more item, the tactic is inert for you — which the Stage 1 analysis reveals in an hour.
Are AI product recommendations enough?
Generic "you may also like" widgets underperform affinity-derived placements because they optimise clicks, not order composition. The co-purchase matrix is your data; use it.
Which agents run this?
Analyst does the derivation and measurement, Builder ships thresholds, bundle pages, and placements, Retention runs the post-purchase extension. It is a standing programme, not a one-off project.
Keep reading
Post-purchase upsell
Raise repeat rate and order value in the window after purchase, without souring the experience.
Design a loyalty programme
A loyalty programme designed from repeat-purchase economics — not points for their own sake.
Merchandising automation
Collection order, featured products, and promotions driven by stock and performance, not by memory.
Improve store page speed
A field-data-first speed programme: measure real users, fix by layer, prevent regression.
Put the team to work on your store
Connect your storefront, analytics, and email stack, set a goal, and let the agents run the work end to end. Start on the free plan with your own model key.