A gain in conversion rate is worth more than the same gain anywhere else, because it applies to every visitor you already paid for and it costs nothing extra to serve them. That is the argument, and the arithmetic behind it is simple enough to do on the back of an envelope with your own numbers.
Start with the equation the whole business runs on. Visitors multiplied by conversion rate multiplied by average order equals revenue. Three of those four numbers are inputs you can move, and the fourth, repeat purchase, decides how many times you get to run the equation for the same customer.
Most stores spend almost all their attention on the first number. It is the visible one, it has an invoice attached, and there is always someone willing to sell you more of it. But visitors is also the only one of the four that costs money to increase. Conversion is bought once and then keeps applying, which changes the order in which we would spend a fixed amount of effort.
The arithmetic, with a worked example
Say a store gets 40,000 visitors a month, converts at 1.5 in every hundred, and has an average order of $70. That is 600 orders and $42,000 a month.
Now move the conversion rate to 2 in every hundred, the same traffic, the same average order. That is 800 orders and $56,000 a month. The store found $14,000 a month, or $168,000 across a year, without buying a single extra visitor.
To get the same result through traffic alone at the original conversion rate, the store would need about 53,000 visitors instead of 40,000. Those 13,000 extra visitors have to be bought or earned, and the paid ones carry a cost that scales every month, forever.
Run the same exercise at your own numbers. A store at 200,000 visitors finds proportionally more. A store at 5,000 visitors finds proportionally less, which matters because it tells a smaller store something important: below a certain volume, conversion work has a real but modest payoff, and traffic is genuinely the constraint. The point is not that conversion always wins. It is that you can calculate which one wins for you in about ten minutes.
Conversion is bought once and then keeps applying to every visitor you were already paying for.
Why the gain compounds instead of sitting still
The revenue figure above understates the result, because a conversion improvement changes the economics of everything else you do.
Your cost per acquired customer falls immediately, since the same ad spend now produces more orders. That means campaigns which were marginal become viable, and you can bid into auctions you previously could not afford. The traffic lever gets easier because the conversion lever moved.
Your break-even return on ad spend falls for the same reason, which widens the range of products you can advertise profitably and lets you compete on terms that used to be out of reach.
And every extra order enters your repeat purchase machinery. If a share of first-time buyers come back, the additional orders this month become additional customers next quarter, each with their own second and third order.
This is why we look at conversion before we look at budget in almost every plan we write. Not because traffic does not matter, but because raising conversion first makes every dollar of traffic spend afterwards work harder.
Where a gain of this size usually comes from
Moving conversion from 1.5 to 2 in every hundred sounds like a redesign. It usually is not. It is normally three or four unglamorous fixes stacked on top of each other, each too small to be exciting on its own.
Shipping cost and delivery date shown on the product page rather than at checkout. This alone changes the shape of the checkout funnel in most stores we look at, because it moves the bad news to a point where the shopper has not yet invested effort.
The two or three questions your support inbox answers every day, written into the buying area of your best-selling pages.
A checkout with fewer fields, guest checkout available, wallets enabled, and address autocomplete working. Every removed step is a small share of people who no longer drop out.
A product page that loads quickly on the mid-range phone your customers actually use, on a mobile connection, not on office wifi.
None of these need a new theme. They need someone to walk the store as a buyer, list what is wrong, and fix the list in order of how much traffic each page receives.
Measure it on the pages you changed
Store-wide conversion rate is a noisy number. It moves when your traffic mix moves, when a sale ends, when a channel changes, and when the season turns. If you judge a product page fix by the store-wide figure, you will usually conclude nothing happened.
So we measure narrowly. If the change was on ten product pages, measure add-to-cart rate on those ten pages, against the same pages in the previous period, with traffic source held as steady as you can.
If the change was in the checkout, measure the step it touched. Cart to checkout, or checkout to purchase, not the whole funnel.
Split mobile from desktop always. They behave differently enough that a blended figure can hide an improvement on one and a regression on the other.
And give it enough volume to mean something. A page with 200 sessions a month will not tell you anything in a fortnight. If your volume is low, change several things at once, accept that you will not know which one worked, and judge the batch. Certainty is a luxury that traffic pays for.
When the answer really is more traffic
We are not arguing that conversion is always the first job. There are stores where it plainly is not, and pretending otherwise wastes a quarter.
If your traffic is very low, the arithmetic above produces a small number and no amount of page work changes that. A store with a few thousand sessions a month needs demand before it needs optimisation.
If your conversion rate is already strong for what you sell, and the funnel shows no obvious drop, then the constraint is elsewhere. Look at order value and repeat rate before you grind for another fraction of a point.
If your traffic is mostly the wrong people, conversion work will not fix it. A campaign sending broad, cheap clicks to a niche product produces a low conversion rate that is a targeting problem wearing a conversion costume.
The way to tell is to run the equation both ways. Model what a realistic conversion gain is worth, then model what the same effort spent on traffic would cost and return. Whichever number is bigger is where the quarter goes. That calculation is the first thing we do when we write a plan, and we show the working.