The four lever model: the complete guide

Write the equation down, because seeing it changes how you spend. Visitors multiplied by conversion rate gives you orders. Orders multiplied by average order value gives you revenue from first purchases. Add repeat purchases and you have the revenue the business actually produces over a year.

Because the numbers multiply, a change in any one of them changes the total by the same proportion, whatever its starting size. Doubling your traffic and doubling your conversion rate have identical effects on revenue. That sounds obvious written down, and yet most store budgets go almost entirely into the first number, because traffic is the one you can buy directly.

The difference is cost. Buying more visitors costs money every month and stops working the day you stop paying. Improving conversion applies to the visitors you already pay for, and keeps applying. Raising average order raises the value of every future order without raising the cost of getting it. Improving repeat purchases means revenue from customers you have already paid to acquire. Three of the four levers make your existing spend work harder. Only one of them requires more spend.

Here is a worked example. Say a store gets 10,000 visitors a month, 2 in every 100 buy, the average order is 60, and a third of customers place a second order during the year. That is 200 orders a month at 60 each. If you want to add half again to revenue, you can buy 5,000 more visitors every month, forever, or you can raise the conversion rate to 3 in 100, or raise the average order to 90, or work on the number of customers who come back. Those routes cost wildly different amounts and take different lengths of time, and only the arithmetic in your own store tells you which is realistic.

The model is not a theory about marketing. It is a way of deciding what to do next.

Pull the four numbers for your store before making any decision about them. They exist in your store admin and your analytics already, and the exercise usually takes an hour.

For visitors, use sessions, and split them by source: paid search, shopping, paid social, organic search, email, direct and referral. One blended traffic number hides everything useful, because the sources behave completely differently and cost completely different amounts. Note the trend over the last year as well as the current month.

For conversion rate, take orders divided by sessions, then split it two ways: by device and by source. The device split matters because most stores get most of their visitors on a phone, and a large gap between phone and desktop is one of the most common leaks there is. The source split matters because traffic that arrives from a search for your brand converts very differently from traffic that arrives from a video.

For average order value, take revenue divided by orders, and look at items per order alongside it. Also look at where your orders cluster relative to your free shipping threshold, because a pile of orders sitting just below it means the threshold is in the wrong place.

For repeat purchases, take the share of customers who have ordered more than once, and the typical time between first and second order. If you sell something consumable, that gap is the single most useful number you have, because it tells you when to send a reminder.

Write the four down with the date. Then work out what each one would need to be for the business to hit its target, and you will usually find that one of them is doing far more work than the others to hold the total down.

Decide which of the four is holding the business back before you spend anything, because the same work applied to the weakest number produces far more than it does anywhere else. The way to find it is to compare your store against itself and against the shape of the funnel, not against an industry figure.

The funnel tells you most of it. Take sessions, then product page views, then add to carts, then checkouts started, then orders. At each step, work out what share moves forward. The step with the sharpest fall is where you are losing people, and it is usually obvious once the numbers are in a column.

Then read the signs for each lever. Visitors is the weakest lever when orders stop the moment ads pause, when almost all traffic comes from one increasingly expensive source, or when your product and category pages get no search traffic at all. Conversion is weakest when traffic is fine but add to cart rates are low, when carts die at the shipping or payment step, or when phone conversion is far below desktop. Average order is weakest when nearly every order is a single item, when the shipping threshold was set by guesswork, or when small orders lose money once postage is counted. Repeat purchases is weakest when most customers only ever buy once, when your email only ever sends discounts, or when you sell something consumable with no reorder route.

Be honest about which of those describes you. Most stores already know, and spend money on the lever they enjoy rather than the one that is broken.

One caution: the weakest lever is not always the one to fix first if fixing it depends on something else. There is no point driving more traffic to a checkout that fails on phones, and there is no point building repeat purchase flows for a product nobody has bought yet. Sequence matters, and the next chapters take the levers in the order they usually need attention.

Visitors are the first number and the one most stores spend all their money on. More of them only helps when the rest of the store can convert, upsell and retain them, so the work starts with where your visitors come from and what each source actually produces rather than with a bigger budget.

Look at cost per visitor by source against orders, not clicks. A cheap visitor who never buys is expensive. Look at the share of visitors who come back on their own, because stores that grow have a rising share of people arriving without being paid for. Look at how many landings go straight to product pages from search, which tells you whether your category and product pages are doing their job.

The usual first moves follow a sensible order. Fix the product feed before raising the budget, because shopping ads are only as good as the data behind them: titles, images, prices and availability decide which searches you show for. Rebuild campaigns around margin rather than popularity, so products with room get the budget and thin margin products get a ceiling. Make category pages answer the search rather than just listing products, because a page that explains, compares and guides is what ranks and what converts. And bring your existing list back, because email and SMS are the cheapest visitors you will ever get.

When we run ads, they run in your own accounts. You own the account, the history and the data, so if you ever stop working with us everything stays with you. Ad spend goes to the platform, in your account, and we never take a percentage of it.

The honest test for this lever is simple. If your conversion rate, order value and repeat rate are healthy and the only thing limiting revenue is how many people see you, buying more visitors is the right answer. If they are not, more visitors will make the leak more expensive.

Conversion rate is the share of visitors who buy, and it is usually where the quickest return sits, because every improvement applies to all the traffic you are already paying for. Nothing else gives you more orders without costing more per month.

The way to find the problem is to walk the store the way a buyer does, on the phone your buyers actually use. Land from a search, find a product, read what is there, add it to the cart, and go to the payment step. Note every place you had to guess, wait or go looking for something. That list, in order, is the work.

The numbers to watch are the add to cart rate, which is the product page's job, the cart to checkout rate, where surprise shipping costs and forced account creation show up, checkout completion, which is decided by payment options and form length, and the gap between phone and desktop conversion.

The usual fixes are unglamorous and effective. Show shipping cost, the free shipping threshold and the delivery window on the product page, because unexpected shipping cost is the most common reason a cart is abandoned. Answer the questions that stop people: sizing, materials, returns, delivery time, what is in the box, all above the fold on a phone. Cut the checkout down with guest checkout, address autocomplete and wallet payments, since every field removed is a few more orders. And make the product, cart and checkout fast on real phones rather than on a desktop simulator.

Most of this is done without a redesign. Rebuilding a store is occasionally the right answer and it is rarely the first one, because the leaks usually sit on the product page, the cart and the checkout. Where there is enough traffic for a test to mean something, we test. Below that, we make the change, measure before and after, and move on.

Average order value is the number most stores never look at, and it is frequently the cheapest one to improve, because the visitors are already buying. The question is only whether they buy one thing or the two things that belong together.

Watch four numbers here. Average order value tracked by channel, so a change in traffic mix does not hide a real change. Items per order, which shows whether bundles and add ons are working. Orders sitting just below your free shipping threshold, which tells you the threshold is set wrong. And margin per order, because a bigger order that gives the gain back in discounts is not a win.

The first move is usually the free shipping threshold. Set it from your own data, just above your typical order, then measure. A threshold set by guesswork either costs margin on orders that would have happened anyway or does nothing at all.

The second is bundles built from real baskets. Look at what people already buy in the same order and offer those things together, with a small reason to choose the set. This is different from bundling what you want to move, which shoppers see through.

The third is one relevant add on at the cart. A single item that genuinely fits what is already there converts. A wall of unrelated products is ignored and slows the page.

The fourth, where the product allows it, is sizes and quantities that reward the larger choice, with clear per unit value so the shopper can see the logic.

None of this is discounting. Most of the gain comes from thresholds, bundles and add ons that raise the order without lowering the margin. Discounts are the last tool here, not the first. And it works for single product stores too, through quantity packs, subscriptions and accessories.

Repeat purchases are the lever that keeps paying after you stop working on it. A customer who buys again costs nothing to acquire, so every improvement here is worth more than it first looks, and it raises what you can afford to pay for the first order too.

The numbers to watch are the repeat customer rate, which is the share of customers who have bought more than once, the typical time between orders, which sets the timing of every reminder you send, the revenue produced by each automated flow measured separately, and list growth alongside open rate, because a list that grows and gets opened is an asset while one that only ever receives discounts stops working.

Flows come before campaigns. Welcome, post purchase, replenishment and win back run on their own and keep working after anyone stops touching them, which is why they are set up first. A weekly campaign that depends on someone writing it will lapse; a post purchase sequence will not.

Where the product is consumable, a subscribe and save option turns one order into a schedule, and the reorder timing should follow the real gap between orders rather than a round number.

Email that people want to open matters more than send frequency. How to use the product, what is new, what other customers do with it. Discounts kept for the moment they matter rather than sent every week until they mean nothing. A win back that gives a reason to return usually beats a percentage off sent to everybody.

You can do all of this on the email platform you already have. The common tools all handle these flows, and a change of platform is worth recommending only when the current one is genuinely the thing in the way. A points based loyalty scheme is usually a later decision, once flows and subscriptions have done the straightforward work.

Change one lever at a time, and write down what the numbers were before you started. Stores that change five things at once get a result they cannot explain, which means they cannot repeat it or reverse it.

Start with the weakest lever, and inside it, start with the fix with the shortest path to an effect. Most of the value in any lever sits in two or three changes rather than in twenty, and doing those properly beats doing everything partially.

Give each change enough time to be readable. How long depends on your order volume: a store with a few hundred orders a month can read a checkout change within weeks, while a store with fewer orders needs longer or needs to look at an earlier step in the funnel where the numbers are bigger. Decide the window in advance so the answer is not decided by whoever looks at the dashboard on a good day.

Expect the levers to interact, and account for it. Raising the free shipping threshold can lift average order and lower conversion at the same time, and the only way to know whether that was a good trade is to look at revenue per visitor rather than at either number alone. Turning on more traffic changes your traffic mix, which moves your blended conversion rate without anything on the site changing. Whenever a number moves, check what else moved with it.

Write the before figures, the change made, the date and the result somewhere permanent. After a year, that record is the most useful marketing document the business has, because it describes what works for this store rather than what works generally.

And when a lever is fixed, go back to the funnel. The weakest lever will have moved, and the next piece of work is somewhere else.

Put the whole thing on paper before anyone starts work. A written plan is what turns a diagnosis into a sequence of decisions, and it is what lets you judge whether the work was worth it afterwards.

The free plan we write is four pages. Page one is where the revenue leaks: your four numbers as we can see them, which one is weakest, and where visitors drop off on the way to an order. Page two is the first three fixes, in order, with the reasoning for each, what each one involves, roughly how long it takes and what we would expect it to move. Page three is what we would leave alone, which is often the most useful page, because the things that are working should not be touched. Page four is the scoped quote: a base monthly fee for the fixes on page two plus a capped growth share on revenue we add, month to month.

How we write it is deliberately unglamorous. We walk your store like a buyer, landing from a search, finding a product, adding it and getting to checkout on a phone, and we note every place a real buyer would hesitate. Then we walk it like operators: traffic sources, product feed, page speed, shipping settings, checkout steps and what happens after the order. We use the numbers you share, a monthly sales band and your goals, which tell us what is realistic and what to prioritise. No access, no logins.

What is not in it: no generic checklist, no pitch deck, no statistics we cannot stand behind, and no pressure to sign. The plan is yours to keep and to act on with us or with anyone else.

If you do work with us, the price is a base fee plus growth share for the scope in the plan. It does not move with your sales or your ad spend, there is no long contract, and you keep everything we build, including the plan.

Repeat the whole process on a schedule rather than treating it as a one off diagnosis. Once you fix the weakest lever, a different one becomes the constraint, and the store that keeps finding the new one grows steadily instead of in bursts.

A monthly rhythm works for most stores. Pull the four numbers and the funnel steps. Compare them with the same month last year as well as with last month, so seasonality does not fool you. Note what changed and what you changed. Decide which lever is now weakest, and what the next single piece of work is.

Keep the report short. Four numbers, five funnel steps, what moved, what we did, what is next. A long dashboard nobody reads is worse than a short one everybody does, and the point of the exercise is a decision, not a document.

Expect the answer to move around. A store that fixes its checkout will often find that average order is now the constraint. A store that gets repeat purchase working will find it can suddenly afford more expensive traffic, which makes visitors the sensible next lever. That movement is the model working, not a sign of drift.

Also re examine your assumptions periodically. Costs change, platforms change what they show and how they charge, and a product mix shifts over a year. The arithmetic that justified a decision in January may not hold in September.

And keep one thing steady: measure the store against itself. Comparing your conversion rate to a figure published about other people's stores tells you very little, because it depends on what you sell and where your traffic comes from. The same store, the same traffic, before and after a change, is the comparison that answers the question you actually have.