eCommerce Growth Strategy: Four Levers, One Constraint

Every growth plan is one of four levers. Working out which one is actually stuck is most of the strategy, and the part usually skipped.

Muhammad Shehryar2026-09-185 min read
Example screens on two phones: a made-up store's pages, of the kind this work is about
Revenue is traffic times conversion times average order value, times how often somebody comes back. There are only four levers, and every tactic anybody sells is one of them wearing a different name.
That is not a simplification for its own sake. It is useful because at any moment one of the four is the constraint, and effort spent on the other three produces very little.

Finding the Constraint

Take the last quarter and write down all four numbers. Then compare each against what the category can support, not against last year.
A store converting at half the category norm has a conversion problem, and buying more traffic into it is buying more of the same disappointment. A store converting well with flat sessions has a traffic problem. A store with strong first orders and no second ones has a retention problem, and no amount of acquisition fixes it.
The mistake that costs a quarter is deciding the constraint from the channel somebody is most comfortable in.

Reading the Four Numbers Honestly

Each number has a source, and each source has a way of misleading. Sessions come from analytics and include visits that were never going to buy: staff, bots that slipped through, people checking an order. Conversion is orders over sessions, so anything that inflates sessions makes conversion look worse than it is. Order value is revenue over orders, and a few large wholesale orders through the store can lift it for a quarter without anything having changed for retail customers.
Repeat purchase is the hardest to read and the most often skipped. The useful measure is the share of a year's customers who bought again within that year, read for customers acquired in the same period, because a store that acquired heavily last quarter will look as though its repeat rate fell when it simply has more new customers who have not had time to come back.
Read each channel separately before reading them together. A marketplace and a direct store have different conversion rates, different order values and very different repeat behaviour, and a blended number describes neither. The constraint is often obvious in one channel and invisible in the average.

One Store, Diagnosed

Illustrative figures for one direct store over a quarter, to show how the constraint is found and why it matters which one is chosen.
The four numbers. 40,000 sessions. A conversion rate of 1.2 percent, which on reading comparable stores is about half what the category supports. An average order of $45. And 15 percent of the quarter's customers buying again within it.
First orders. 40,000 sessions at 1.2 percent is 480 orders, at $45 each about $21,600.
The option the team wanted. More advertising, buying 20 percent more sessions. At the same conversion that is 576 orders and about $25,900, a rise of a fifth, paid for in advertising on every extra visit.
The option the numbers point to. Conversion is the number furthest from what the category supports. Lifting it from 1.2 to 1.8 percent, still below the norm, turns the same 40,000 sessions into 720 orders and about $32,400: half as much again, with no extra traffic bought, and every future visit converting better too.
What the fix looked like. The objections were in the support inbox and the reviews: delivery times nobody could find before checkout, a returns policy buried in the footer, and sizing questions answered nowhere on the product page. Each fix was a sentence or a section, not a redesign.
Then the constraint moved. With conversion near the norm, traffic became the constraint, and the advertising the team wanted in the first place became the right next step. The order is what made it pay.

Traffic, When It Is the Constraint

Three sources, and they behave differently. Marketplace demand exists already and costs a referral fee and advertising. Search compounds and is slow. Paid social creates demand and stops the day the spend does.
The right answer for most product businesses is not one of them. It is the marketplace first because the demand is there, then search because it compounds, then paid to accelerate what is already working.

Conversion, Which Is Usually Cheapest

A conversion problem is the cheapest of the four to fix, because the traffic is already paid for. It is also the one most often diagnosed wrongly: the page is blamed when the problem is the offer, the delivery promise or the price against the alternatives.
Look at the objection rather than the layout. Reviews, support tickets and the questions on marketplace listings say what stops the purchase, and almost none of them are solved by moving a button.

The Conversion Fixes That Usually Come First

When conversion is the constraint, the same handful of causes turn up in store after store. Checking them in order is quicker than redesigning anything.
The delivery promise. Shoppers want to know when it arrives and what delivery costs before they commit. A date or a range on the product page, and the shipping cost shown before checkout rather than at the last step, removes the most common reason a basket is abandoned.
The returns policy. A clear, findable policy near the buy button lowers the risk a first time buyer is taking. Hiding it does not reduce returns; it reduces orders.
The answer to the question every buyer asks. Sizing, compatibility, what is in the box, how long it lasts. The question that fills the support inbox is the sentence missing from the product page.
Proof that others bought it. Reviews on the page, from real customers, with the critical ones visible. A product with no reviews converts like a product nobody has tried.
The price against the alternatives. Shoppers compare in another tab. If the same product is cheaper on a marketplace, the store's own page loses the sale however good it is.
The checkout itself. Forced account creation, unexpected costs at the last step, and payment methods the customer does not use. Each is a reason to leave at the point of buying.
Fix them in that order, one at a time, and read conversion for a few weeks after each. Most stores find the constraint has moved before they reach the end of the list.

Order Value, Which Is Quietly the Fastest

Bundles, multipacks, and the honest version of an upsell, which is offering the thing the customer would have bought anyway if they had thought of it.
This lever gets ignored because it feels less like growth than new customers. It is frequently the fastest of the four, because it requires no new traffic and no new trust.

Repeat Purchase, Which Decides What You Can Afford

Repeat rate is what sets the acquisition cost the business can carry. A brand with genuine repeat purchase can outbid one without it for the same customer and still make more money.
It is also the lever a marketplace makes hardest, because the buyer belongs to the marketplace rather than to you. That is the real argument for a direct channel, and it is an argument about economics rather than about brand.

How the Levers Pull on Each Other

The four levers multiply, so moving one usually moves another the opposite way. New traffic from a channel the store has not used before converts worse at first, because it is colder: conversion falls while revenue rises, and a team watching conversion alone will call a working campaign a failure. Bundles and multipacks raise order value and can lower conversion, because a larger basket is a bigger decision.
Discounting is the clearest case. It lifts conversion, lowers order value and margin, and teaches returning customers to wait for the next one, which quietly changes the repeat lever too. A quarter of growth bought with discounts can leave the business with more customers and less profit per customer than it started with.
So the measure of any change is the product of the four and the margin left after it, not the lever it was aimed at. A change that lifts its own number and lowers revenue, or lifts revenue and lowers margin, has moved the problem rather than solved it.

The Plan That Survives Contact

One constraint, one quarter, one measurement that says whether it moved. Anything wider than that becomes a list of activities nobody can evaluate.
Write down what would make you stop. A plan with no stopping condition is a plan that runs until the budget does.
The channels are means rather than the plan. What we run and how it fits together is on the services overview, and the honest way to judge any of it is against margin rather than against platform metrics.

A Quarterly Review on One Page

The review that keeps a plan honest fits on a single page and takes an hour. The four numbers for the quarter beside the quarter before, by channel. The constraint chosen at the start, and the evidence for it. The one change made to move it. The measurement that was meant to say whether it moved, and what it said.
Then the stopping condition written at the start, and whether it was met. If it was, the change is stopped, whatever anybody feels about it, and the page says what was learned. If it was not, the page says whether the constraint is still the same one, because a constraint that has been relieved moves somewhere else, and next quarter's plan starts from wherever it has gone.
Kept every quarter, the pages become the most useful document the business has: a record of what was tried, what it cost and what it did, which is exactly what is missing when a new channel or a new agency is being considered and everybody remembers the past differently.

Building Repeat Purchase on a Store You Own

Repeat purchase is the lever a direct store controls best, because the store knows who its customers are. Most stores pull it with discounts, which is the most expensive way to pull it.
A useful first message after the order. How to use the product, how to care for it, what to expect. A customer who gets more from the first purchase is the one most likely to make a second.
Reminders timed to the product. For anything that runs out, a reminder shortly before it would run out, based on how long the product lasts. It is the most natural repeat purchase there is, and it needs no discount.
Subscriptions where they fit. For consumables bought on a regular cycle, an optional subscription removes the effort of reordering. It suits products people already rebuy; it does not suit products people buy once.
Recommendations that make sense. The accessory, the refill, the next size, offered because it goes with what the customer bought, not because it is in stock.
Rewards without training customers to wait. Early access, a gift with a later order, or service benefits reward loyalty without teaching customers that the price always drops if they hold off.
Measure by group. Track repeat purchase for customers acquired in the same month, so a quarter of heavy acquisition does not make the repeat rate look as if it fell.
None of this is available for marketplace customers, whose details belong to the marketplace. It is the strongest reason a brand selling on marketplaces still builds its own store.

What We Do With This

We are asked for a channel and we start with the constraint, because a brand whose conversion is the problem does not need more Google Ads or more Meta, it needs the pages fixed first.
Where the constraint is genuinely traffic, the marketplaces are usually the fastest route: Amazon, Walmart and TikTok Shop have the demand and charge for access to it.
If you want the four numbers read against your own quarter, that is where we start.

Frequently Asked Questions

The questions that come up most often on this subject.

Want This Applied to Your Account?

Tell us the category and the numbers you have. We will review the account and tell you where the opportunity is and what we would change first.