Raising average order value is the cheapest growth lever you have, because none of it costs more ad spend. It is also the easiest place to congratulate yourself on a number that made you poorer. A store that pushes AOV from ₹1,850 to ₹2,150 with a 20% bundle discount has moved the headline metric and cut gross profit per order. So before any of the tactics below, decide what you are actually optimising: contribution margin per order, after discount, after shipping, after the returns you will eat.
Your AOV is one number hiding a distribution
Open Shopify Analytics, export orders for the last 90 days, and build a histogram of order totals in ₹250 buckets. Almost every store we look at has a shape nobody in the business has seen. Usually there are two or three spikes: single-unit orders at the hero product's price, a cluster just above the free shipping threshold, and a thin long tail.
That shape tells you which lever applies. If 60% of orders are one unit of one SKU, bundling and multi-buy are your play and threshold games will do nothing. If orders are already 2.4 units on average and clustered tightly around the shipping threshold, you have a pricing-ladder problem, not an attachment problem. Also check the median against the mean. When the mean sits 30% above the median, a handful of large B2B-ish orders are flattering your dashboard and every A/B test you run on AOV will be noisy.
One reporting trap: Shopify's reported average order value is built from total sales, which includes shipping and tax. Change your shipping fee and AOV moves without a single extra product being sold. Pull product revenue divided by orders separately and track that as the honest version.
Setting the free shipping threshold with arithmetic instead of instinct
The most common request we get is "should we raise free shipping from ₹1,499 to ₹1,799?" It is answerable. Here is the shape of the calculation on a store with 55% gross margin and ₹90 average shipping cost per order.
Say 22% of orders currently land between ₹1,499 and ₹1,798 — the band that loses free shipping when you raise the bar. Per 1,000 orders that is 220 affected. Of the customers who do not walk away, assume two thirds add roughly ₹300 to cross the new line and one third pays a ₹79 shipping fee instead.
- Adders: 220 × (2/3) × ₹300 × 0.55 margin = about ₹110 of extra gross profit per affected non-abandoning order
- Shipping payers: 220 × (1/3) × ₹79 = about ₹26 per affected non-abandoning order
- Abandoners: each one costs ₹1,650 × 0.55 = ₹907 of gross profit, less the ₹90 shipping you no longer pay, so ₹817
Set the two sides equal and the break-even abandonment rate in that band works out to roughly 14%. If more than about one in seven of those customers leaves rather than pay ₹79 or add another item, the raise loses money — and it loses it quietly, because your AOV chart goes up while profit goes down.
Fourteen percent is not a comfortable margin of safety. Run it as a real test with a holdout, watch orders per session in that band, and give it three weeks. In India, where a ₹79 shipping line is a genuine objection for a large share of buyers, we have seen the raise fail more often than it works. The version that does work: keep the threshold, and put a progress indicator in the cart drawer that shows the exact rupee gap plus two suggested add-ons priced to close it.
COD makes every AOV win provisional
Push AOV up on a COD-heavy store and you have also pushed up the value of every parcel that comes back undelivered. If your RTO rate on COD is 18% and you raise COD basket sizes by ₹400, you are now writing off more per failed order plus the return leg freight, and the cash sits with the courier for another two weeks.
The fix is to make the AOV growth prepaid. Concretely: cap COD at a ceiling — a lot of Indian brands sit somewhere between ₹2,000 and ₹5,000 depending on category — and put your upsell mechanics behind prepaid. A 5% prepaid discount looks expensive until you price it against RTO. On a ₹2,000 order at 55% margin with ₹90 forward and ₹90 return freight, a failed COD delivery costs you roughly ₹180 in freight plus handling and re-stocking, before the working capital hit. Converting even a fifth of your COD orders to Razorpay or UPI usually pays for the discount several times over.
Partial COD — collect ₹100 to ₹200 upfront via UPI, rest on delivery — is the underrated move here. It filters out the impulse orders that never get accepted, and it does not require the customer to trust you with the full amount.
Bundles that hold their margin
Three bundle types, in descending order of how well they work in our experience:
Fixed-price sets with their own SKU. A skincare routine, a three-piece cookware set, a gifting box. The customer never sees the component prices side by side, so you are not visibly discounting. You can build these as real products with their own inventory, or use Shopify's native bundles so components draw down individually. Margin stays where you set it.
Volume tiers on consumables. Buy 2 save 8%, buy 3 save 12%. Works for supplements, coffee, pet food, anything with a refill cycle. The trick is making the middle tier the default selection on the product page, not the single unit. Changing the pre-selected variant is a twenty-minute theme edit and it reliably moves units per order more than any banner.
Complete-the-look cross-sells. Highest ceiling, most work. This only functions if the recommendations are genuinely curated. Shopify's default related-products logic pulls from the same collection and will happily show a customer buying a size-M shirt three more size-M shirts. Hand-map the top 30 products to two accessories each. That covers most of your revenue and takes an afternoon.
What consistently underperforms: gift with purchase where the gift is obviously stock you could not sell. Customers read it correctly.
A quick GST caution on bundling
If you merge items taxed at different rates into one SKU sold at a single price, GST treats it as a mixed supply and the whole thing can attract the highest applicable rate. Apparel has its own wrinkle, since the slab depends on per-piece price — bundle two lower-priced garments into one higher-priced listing and the tax treatment can change. Talk to your CA before you build the SKU, not after your first quarter of filings.
Post-purchase upsells beat pre-purchase almost every time
An offer on the thank-you page cannot hurt checkout conversion, because checkout already happened. That asymmetry is why we start there. Attach rates in the 4-10% range are normal for a well-chosen single-product offer; anything above that usually means the product was underpriced in the first place.
Two things to know. The post-purchase page does not render for every payment path, so measure attach rate against orders that were actually shown the offer, not total orders, or you will conclude it failed when it did not. And editing the checkout page itself — an upsell between shipping and payment — is a Shopify Plus capability. If that is the only reason you are looking at Plus, it is not enough of a reason; we say so regularly to brands considering the upgrade. Get the cart drawer and thank-you page working first.
Keep the post-purchase offer to one product, one price, one click. Two offers in sequence cuts total take-rate in most tests we have run.
On big catalogues, discoverability is the real constraint
Stores with 800+ SKUs rarely have an upsell problem. They have a findability problem. The customer bought one thing because that was the one thing they could find. When filters are limited to a single generic dropdown and site search returns nothing for a plural or a misspelling, the second item never enters the cart.
Look at your search terms report. If more than 15% of searches return zero results, fix search before you build a single bundle. Better faceted filters — material, price band, occasion, in-stock only — routinely lift units per order because they let a customer who came for one category browse a second. We built FilterPro for exactly this after doing it by hand on too many theme builds.
Things we would skip
- Spin-to-win popups. They discount to people who were already buying.
- Threshold nag bars that appear on every page. Cart and product page is enough.
- Loyalty points as an AOV lever on a store doing under a few hundred orders a month. The maths does not clear the app cost.
- Launching any new AOV mechanic in the first week of October. Test in a quiet month; ship the winner before the festive traffic arrives, not during it.
Measure it properly or don't bother
Track four numbers weekly, side by side: units per order, product revenue per order, gross profit per order, and RTO-adjusted gross profit per order. A tactic that raises the first three and drops the fourth is a tactic that shifted your mix toward high-value COD. That happens more than people expect on Indian stores, and the dashboard will not tell you unless you build the column.
Also watch page weight. Upsell widgets, bundle builders and recommendation apps each add scripts, and a cart drawer that takes 900ms to open loses more revenue than the upsell inside it earns. We have removed AOV apps from stores and watched revenue go up.
Where to start
Pull 90 days of orders, build the histogram, and find the two spikes. Then pick the one lever the shape argues for — pre-selected multi-buy, a curated cross-sell map, or a post-purchase offer — and run it against a holdout for three weeks. One change, measured properly, beats six shipped at once.
If you want a second pair of eyes on the distribution and the threshold maths, our free store audit covers it, and if the answer turns out to be theme work rather than another app, that is what we will tell you. Teams that would rather run it in-house can bring in a developer on a short engagement for the cart and product page changes.