Same Orders, Half the Ad Spend | The Discoverability Company

Case studies / Same orders, half the ad spend

Same orders, half the ad spend.

A household-cleaning brand on Shopify had run Google Ads through an outside agency since 2017, quietly losing money on it. We took control of the account and cut the media budget by more than half without losing a single order per month.

54% Less media spend, same order volume
2.1x Better cost per completed order
0.80x to 1.94x Tracked return on ad spend (2024 vs May-Jun 2026)
54% less media spend at the same monthly order volume

The short version

We don't name clients. Here's exactly what happened anyway. In 2024 this account produced about 28.5 orders a month on $1,707.01 of monthly spend. In the first half of 2026, under our control, it produced about 28.1 orders a month on $791.62. Same output, 46% of the money, and the channel went from losing money to making it.

Research: what we found

The product is a household cleaner. About $24 an item, average order somewhere in the thirties or low forties. An outside agency had held the Google Ads account since 2017, and by 2024 it was spending $1,707.01 a month to bring back 80 cents on the dollar.

Nobody catches this one. There is no blowup to point at, no month where the wheels come off and somebody calls a meeting. Just a small loss, over and over, that looks unremarkable on any single report you happen to open.

There was also a measurement problem, and it was the reason nobody had caught the first one. The conversion reporting counted add-to-cart and begin-checkout events as conversions, right alongside completed orders. Take the worst month. It reported 672 conversions. Of those, 664 were cart-adds and 3 were checkout starts. Five were orders. The dashboard put that month at a 13x return. It was 0.31x.

So the number everyone was watching counted browsing as buying, and it had been doing that for years.

Plan: what we told them

We put it to them in plain terms: the account did not have a spending problem, it had a counting problem, and until that was fixed no budget change could be trusted. We told them more spend would not work. Scaling a channel that optimizes toward cart-adds buys more browsing, at a higher price. We also told them the fix would make the reported numbers look worse before they looked better, because hundreds of phantom conversions were about to disappear from the dashboard.

The plan they approved: fix the measurement first, then cut spend to what the real numbers supported, then rebuild from there. In that order, with completed orders as the only score that counted.

Execute: what we did

1
Fixed what counted as a sale

Before touching a bid, we separated completed purchases from cart-adds and checkout starts. Optimizing toward a metric that counts browsing builds an account that is excellent at generating browsing. Every decision after this point was measured against completed orders only.

2
Cut the budget hard, then held it

We rebuilt the account and cut the broad search terms spending without converting. Monthly spend went from $1,707.01 to $791.62, and stayed there rather than drifting back up once the account looked stable.

3
Stopped paying twice for the same customer

Branded searches are the cheapest traffic any business has, and they were being bought inside campaigns built to find new customers. We split brand into its own campaign so the prospecting budget stopped paying premium prices for people who already knew the brand.

4
Worked it every week

Search terms reviewed and negatives added. Ad copy rewritten and tested. Budget moved toward whatever converted that month. Campaigns that stopped earning their keep paused rather than left to drain. None of it dramatic. All of it every week.

Media spend

The first job was spending materially less without losing the orders the account was already producing.

Cost per completed order

Spending less normally costs you efficiency, because the cheap volume is the first thing to go. Here it moved the other way.

The results

The account, before and after control
  • Average monthly spend
    $1,707.01 $791.62 Improved
  • Orders per month
    28.5 28.1 Held
  • Cost per completed order
    $59.90 $28.14 Improved
  • What counted as a conversion The measurement fix
    Cart-adds folded in Orders only Improved
  • Tracked return on ad spend
    0.80x, losing money 1.55x half / 1.94x May-Jun Improved
Completed purchases only, on both sides

Monitor: what we watch now

We look at this account every day, and we push adjustments every single week. That cadence is the product. Most ad accounts get opened when something looks wrong; this one gets opened before anything does, which is how small drifts get caught while they are still small.

The weekly work is the unglamorous kind that compounds. New ad creative written and tested against what is already running. New audiences tried against the ones that carry the account. Budgets shifted between campaigns toward whatever converted that week, and away from whatever stopped. Search terms reviewed and negatives added. The completed-orders definition re-checked so the counting problem never creeps back in. It is the ongoing work most business owners know they should do and do not have time for, done on a schedule that does not slip.

The numbers on this page come from live reads of the account, not from a launch-week snapshot. The most recent full re-pull of every figure here was July 20, 2026, and the account was holding at 1.94x tracked return over its two most recent complete months.

The account held at the new spend level.

If this sounds like your account

The tell is a dashboard full of conversions that never quite turns into money in your bank account. Send us your numbers and we will tell you what is actually being counted, or read how our advertising management works.

Think your ad account is quietly losing money? Let us look at what is actually being counted.

Every conversation is confidential. Tell us what shows up when someone searches you, and we will tell you straight what is possible.

Start the conversation

Prefer email? Write to us directly.

Methodology and confidentiality