Search how to increase average order value and the first page is a payments company, a customer relationship platform, a banking startup and a billing vendor. The lists are fine. Sixteen ways, twenty-eight tactics, eleven strategies.
None of them has had to install one, watch the take rate come back under 10%, and work out which of the four things they changed caused it.
That is the part this covers. Not the list of levers, the settings you point them at and the check that tells you which one is broken.
Key Takeaways
- There is no credible universal average order value benchmark. Your own trend against your own baseline is the only comparison that describes your store.
- Every offer has to survive one question: why does this make sense right now? If the answer is weak, do not show it.
- A free shipping threshold set 15 to 40% above your current average, with a live progress bar, is the single highest-leverage lever.
- Relevance beats discount. A relevant offer at 10% off usually outperforms a random one at 50% off.
- Take rate under 10% with a flat average means the architecture is wrong, not the offer.
Start by refusing to benchmark yourself
The most common opening question is "what should my average order value be?" There is no honest answer to it. Category, price point and traffic source move that number far more than store quality does, and a $40 supplement brand and a $400 furniture brand cannot share a target.
So the useful version of the question is narrower. Given this basket, what is the next thing a buyer would plausibly want, and where in the journey would they want it?
That reframe is what turns a list of tactics into a system, and it is the thing the tactic lists skip.
The one question that kills most upsell ideas
Before any offer goes live, it has to answer one question:
Why does this make sense right now?
If the answer is weak, do not show the offer. The customer should think yeah, that actually makes sense. The moment they think they're trying to squeeze more money out of me, you have spent trust to earn a few dollars, and trust is the more expensive of the two.
The priority order I apply, in this sequence: relevance, then timing, then simplicity, then margin protection, then customer experience. Most stores optimise that list backwards, starting with margin and ending with relevance.
Relevance itself has a hierarchy, and it is worth being blunt about the bottom of it:
- Highest. The offer complements, improves, protects, extends or replenishes what they just chose.
- Medium. Related category, bundle expansion, a genuinely similar product.
- Weak. Trending products, unrelated clearance, generic bestsellers, anything picked to move inventory.
That last row is where most automatic recommendation widgets live. They are chosen by an algorithm optimising for what other people bought, which is not the same question as why does this make sense right now.
Where the revenue actually is
These are the placements worth installing, with the defaults I set them to and the ranges I treat as healthy. Read the right-hand column as typical healthy ranges from the stores I have worked on, not as promises. What you get depends on your traffic, your offer, your price point and your margins.
| Placement | What it is for | Defaults | Typical healthy range |
|---|---|---|---|
| Product page bumps | Low-friction add-ons near the buy box | 1 to 3 max, priced at 5 to 25% of the product, 10 to 20% off | 15 to 40% attach, 5 to 15% of average order value |
| Below-cart-button cross-sells | Offers that need a sentence of explanation | 1 to 2 max, framed as "works best with" | 5 to 15% click rate, 3 to 10% attach |
| Free shipping threshold | The top average order value lever | Set 15 to 40% above current average, live progress bar | 10 to 30% lift, 20 to 60% of eligible customers reaching it |
| Cart reward ladders | Gamified basket building | 2 to 3 tiers, escalating value | 10 to 25% lift |
| Cart upsells | Strongest pre-checkout intent | 1 to 2 impulse items at 10 to 30% of cart value | 5 to 20% attach, 5 to 15% cart value increase |
| Post-purchase | Highest leverage, no checkout risk | 1 offer, 20 to 40% off, sweet spot 25 to 30% | 8 to 20% acceptance, 20%+ exceptional |
Two things to notice about that table.
The post-purchase row is the only one that cannot cost you an order. It fires after the transaction confirms, so there is no abandonment risk to trade against the upside. Everything above it is a bet placed while the sale is still in play. I build post-purchase systems through Skuology and Upsellr, both of which I own, so treat that preference as a disclosed one rather than a neutral observation. The reason still holds independently: you cannot break a checkout you are no longer in.
The ranges are wide on purpose. A 15% attach rate and a 40% attach rate are both inside "working." If someone quotes you a single number for any of these, ask which store it came from.
The threshold, which is the one most stores set wrong
The free shipping threshold is the highest-leverage single change available to most stores, and it is usually set at a round number somebody picked in 2021.
Set it just above your current average order value, at the nearest psychological price point: $42 becomes $59, $68 becomes $79, $83 becomes $99. The useful gap is roughly 15 to 40% above where you are now. Below that it is free money you were already giving away. Above it, most carts never reach it and the bar becomes decoration.
Then show the progress. A static "free shipping over $75" line is a rule. A live "you're $12 away from free shipping" is a goal, and a goal changes basket composition in a way that a rule does not.
One detail that costs stores money quietly: if you tell someone they are $8 from free shipping and then give them no fast way to find an $8 item, you have created a problem and withheld the solution. Solve it at the moment you raise it. That is the whole argument for the cart drawer doing more than listing what is already in it.
Rewards are not interchangeable either. Ranked best to worst by what actually moves baskets: free shipping, then free gifts, then bonus quantity, then store credit, then discounts. A ladder that runs $50 free shipping, $75 free gift, $100 for 10% off gives you three reasons to add rather than one.
The deeper version of this, including when a threshold is the wrong lever entirely, is in the free shipping threshold post.
Relevance beats discount, and it is not close
The instinct when take rate is low is to raise the discount. It is the wrong first move almost every time.
A highly relevant offer at 10% off usually beats a random offer at 50% off. The shopper is not evaluating whether your deal is good. They are evaluating whether the product makes sense for them, and no discount makes an irrelevant product relevant. It only makes it cheaper to regret.
Pairing rules that hold up: the post-purchase offer should improve the first purchase. Toothbrush to toothpaste. Protein to shaker. Shoes to insoles. Coffee machine to pods. Frame it as exclusive and tied to the order that is already placed, because that is true: add this before it ships.
The two results I can point at both came from getting relevance right rather than from discounting harder. An apparel brand added 50%+ to average order value in 30 days on the same products and the same traffic. A pain-relief brand moved from $20 to $39. Neither of those is a promise about your store and neither came from a bigger coupon.
The one I lost
A bedding client wanted mix-and-match bundles where each item in the bundle carried its own size dropdown. I argued it was too complex and would confuse people at the exact moment they were deciding, and I proposed quantity-tiered discounts instead, which are simpler to build and simpler to understand.
The client insisted. I built it, and I mapped every combination of the flow to make sure it held together.
I still think simplicity usually wins at that moment in the journey. But my argument was a general principle applied to a category I had not sold in, and theirs was specific knowledge about how their own category is actually bought. Principles lose to specific knowledge more often than anyone who sells frameworks likes to admit.
The practical lesson is not "complexity is fine." It is that the objection "this is too complex" needs to be tested rather than asserted, especially when the person disagreeing with you sells the product every day.
The diagnostic that tells you what to fix
When something is not working, look at take rate and average order value together, not separately.
- Take rate under 10% and the average has not moved. The architecture is wrong. The offer is in the wrong place, or it is competing with three other offers, or it is answering a question nobody was asking at that point in the journey.
- Take rate healthy and the average flat. The placement works and the offer is priced too low relative to the order it attaches to. Raise the price of the add-on rather than the discount.
- Take rate healthy, average up, margin down. You bought the lift. Check cost of goods before you celebrate, because revenue and profit are not the same win.
That third row is the one that catches people. Choosing revenue over profit is on my list of the most repeated mistakes in this whole category, and it is the easiest one to make because the dashboard only shows you the half that went up.
What to do in the next two weeks
- Measure your current average order value over a clean 30 day window, before you change anything. Without this you cannot tell a lift from a good month.
- Set one threshold 15 to 40% above that number, at a psychological price point, with a live progress bar.
- Install one post-purchase offer that improves the first purchase, at 25 to 30% off.
- Leave everything else alone for 30 days. Two changes you can attribute beat six you cannot.
That is deliberately three changes and not thirty. Launching too many experiments at once is how stores end up with a higher average order value and no idea which lever produced it, which means they cannot repeat it.
If you want the wider system this sits inside, the hub is the average order value playbook. If you are deciding what kind of offer belongs where, cross sell vs upsell draws the line, and the shopify bundle app post covers the bundling side.
If you would rather not spend the next quarter working out the settings yourself, that is what the Invisible Second Sale™ engagement does: the architecture, installed, in three weeks. Book a call and bring your current average order value with you.
