Kovalchick Analytics

Results

What happened on my own brand

I sell physical therapy products on Amazon under PhysioSpace, my own label. That account is where this system was built, it is where every change gets tested before a client ever sees it, and it is the one I can show you the weekly numbers for.

The numbers

Measured per product, per week, after cost of goods, Amazon fees, fulfilment and refunds. Not an ACoS figure.

$83Jul 19$285Jul 26$300Aug 2$256Aug 9$308Aug 16$379Aug 23$432Aug 30WEEKLY CONTRIBUTION PROFIT, $5.2x OVER SEVEN WEEKSafter cost of goods, Amazon fees, fulfilment and returns

Seven weeks, July to August 2026

Weekly contribution profit went from $83 to $432. The turn came after I added two guards: a spend circuit breaker, and a profit floor that refuses to bid above what a product can actually earn.

Before those were in, the first month live was a ramp. One account is one account, so the honest word here is correlated, not caused, and I am happy to walk you through the whole series on a call.

5.2xWeekly contribution profit, July to August 2026
$83 to $432First week of the window to the last
24.4%Total advertising cost of sales across the period
4,181Logged account changes since April, each with a before, an after and a reason

Why one target does not work

This is the part that separates what I do from what a tool does, and it is the reason the numbers above moved.

ONE ACCOUNT TARGET, 3.00xProduct A1.95xProduct B2.33xProduct C3.17xbreakeven return on ad spend, computed from each product’s own cost of goods

Three products, three different bars

These are three real products from one account. They break even at 1.95x, 2.33x and 3.17x. The account was being steered to a single 3.00x target, which is above what two of them need and below what the third needs.

That one number is why 17 of their 20 campaigns were being told to cut or pause, including ten whose return had improved.

Every tool that asks you for one target return on ad spend has this problem. It will cut the bid on your best product and scale the one that cannot pay for the click, and the account level report will look fine while it happens. Getting this right needs your real cost of goods per SKU, which is why I ask for it at intake and why most of the field does not.

What else you could buy

Ordered by what you would pay at $30,000 of spend. My two rows show the founding rate in bold with the standard rate under it. Competitor prices were checked on 2026-09-11 from each vendor's own published pages.

OptionHow it is pricedAt $10k spendAt $30k spendWho runs it
Self serve softwareFlat fee, often plus 2% to 3% of spend$150 to $800$250 to $900You do
Kovalchick Analytics, ExecuteFlat fee, no percentage of spend$649$1,349 standard$649$1,349 standardI do
Kovalchick Analytics, ContinuousFlat fee, no percentage of spend$1,649$2,449 standard$1,649$2,449 standardI do, daily
Typical agency floorFlat retainer, published from $2,000$2,000+$2,000+They do
Ad Badger managed service$975 plus 8% of spend, $2,500 monthly minimum$1,775$3,375They do
Full service agencyPublished benchmark bands$1,000 to $2,500$3,500 to $6,000They do

The shape of that table is the reason I exist. Below roughly $800 a month everything is software you operate yourself. Above roughly $1,700 a month everything is an agency. In between there is one managed service, it takes 8% of your ad spend on top of its fee, and it will not take you at all under about $19,000 of monthly spend.

If you are spending $5,000 to $20,000 a month, your only real alternative to doing this yourself is an agency with a minimum. That is the gap I work in.

At the founding rate, Execute sits inside the price band where everything else is software you have to operate yourself. That is the whole point of the founding window, and it is not going to be true afterwards. The full pricing is here.

See whether the same thing is happening to you

The audit is free and it is the same analysis. Your cost of goods per product, your campaigns scored against their own bar, and the three changes I would make first.