The principle
Every SKU has a job®. Your Google Ads account should reflect it.
One of the first things we want to know when we look at an ecommerce account is: what job is each SKU doing? Not “What ROAS is PMax getting?” We couldn’t give a shit yet. Show us the products.
Campaign ROAS alone cannot tell us whether the right products are selling for the right commercial reasons. Before we celebrate a number, we need to know what the business needs those sales to do.
Same product. Different situation. Different job.
Change the situation and watch the decision change. The product never moves; its job does.
The situation
Proven seller, full size run, contribution holding at volume.
The job: Scale
Give it the next pound. Open targets while contribution per sale holds.
Your business is not made of Google campaign types.
This is the bit we think PPC has got backwards. The agency conversation starts with the campaign types. Then the products are fitted around that structure.
- Search
- Shopping
- PMax
- Demand Gen
Campaign types are delivery tools. They are how ads reach people, not a description of what your products need to achieve. When the account is organised around them, a campaign report can show a healthy ROAS while products with very different margins, stock positions and commercial needs are treated as if they were the same.
A high return on thin-margin products can be a poor trade. A lower return on ageing stock can be a rational way to release cash. The campaign number alone cannot tell you which of those is happening, because it was never designed to. Commercial product priorities have to come first; the campaign structure should serve them.
Start with the products.
Your business is made of products. Each one has a price, a cost, stock on hand and a reason someone might buy it. So the first questions are about them:
- 01
Which acquire customers?
- 02
Which leave contribution after costs?
- 03
Which need to clear ageing stock?
- 04
Which get clicks but fail to convert?
- 05
Which deserve more visibility?
- 06
Which should stop receiving spend?
Every SKU has a job®. Google’s job is to help them do it.
Different jobs need different measures of success.
These are the five BOI® jobs. A product has one primary job at a time. This is a way to ask better business questions, not five new campaign types or five automatic bidding rules.
Staged illustration. Any product can hold any job, and a job can change.
01ScaleGrow worthwhile demand while the economics still hold.Show details for Scale
Grow worthwhile additional demand without losing the economics that make growth valuable. Scaling a product only makes sense while each extra sale still leaves contribution the business is happy to accept.
Ask: Can more demand be won at an acceptable contribution, and how would we know when it stops being acceptable?
Illustrative example, not a client result: A best-selling jacket still earns healthy contribution at its current cost per sale. The decision: raise its budget step by step and set the contribution level at which further growth stops.
02ProfitEarn contribution after the real costs of selling.Show details for Profit
Earn contribution after the real costs of selling and advertising. These are the products that pay for everything else, so their measure is what a sale leaves behind, not the revenue it reports.
Ask: What does this product actually leave behind after product cost, fulfilment, returns, fees and ad spend?
Illustrative example, not a client result: A steady-selling skincare refill reports a modest ROAS but very low returns and fulfilment cost. The decision: keep funding it on what each sale leaves behind, rather than cutting it for a weaker headline ROAS.
03ProtectDefend important demand where paying for visibility is justified.Show details for Protect
Defend commercially important demand where there is a clear reason to pay for visibility. That reason should be stated, priced and questioned, not assumed. Incrementality stays an open question, and Protect is never automatic brand spend.
Ask: What demand needs defending, what evidence supports paying for it, and what would happen if we did not?
Illustrative example, not a client result: A hero product competitors bid on during peak season. The decision: pay for visibility on named terms with a stated reason and a check on what happens without it, rather than defending every brand search by default.
04RecoveryTurn ageing or excess stock back into cash, within limits.Show details for Recovery
Release cash from ageing or excess stock within an agreed contribution floor and time window. The point is turning stock back into money before its value deteriorates. Low stock is a different situation: it constrains spend, it is not a Recovery job.
Ask: Which stock needs to become cash, by when, and what is the lowest contribution we will accept to achieve it?
Illustrative example, not a client result: Last season's homeware range is tying up warehouse space and cash. The decision: give it a time-boxed push with a minimum contribution per sale, then stop once the stock is cleared or the floor is reached.
05GatewayWin customers or baskets where the wider value is measured.Show details for Gateway
Bring in customers or valuable baskets where the wider economics are measured. A Gateway product may accept a weaker first order because repeat purchasing or basket attachment is evidenced, not because a lifetime-value story sounds plausible.
Ask: Is the wider customer or basket value supported by measured repeat or attachment economics, rather than assumed?
Illustrative example, not a client result: A low-priced starter item that first-time customers often buy before returning for larger orders. The decision: accept a thinner first order only where repeat purchasing is measured in the business's own data.
Ignore the jobs and you miss where the money goes.
ROAS is not contribution.
Attributed revenue is not what a sale leaves behind. Contribution before advertising means revenue less product cost, fulfilment and shipping, returns and payment fees. Contribution after advertising deducts the media cost too. Neither figure is automatically net profit, which also has to cover fixed overheads. If those inputs are missing or unreliable, the honest answer is to say so, not to dress a revenue number up as a profit measure.
An attributed sale is not automatically an additional sale.
A product that sells organically does not prove paid spend on it is wasted. Equally, a sale Google reports does not prove the ad created demand that would not have arrived anyway. The useful question is what the advertising adds on top of what would have happened regardless. That matters most for familiar brands and products people were already looking for.
Concentration needs a reason.
Budget concentrated on a few products is not automatically wrong. A small part of the catalogue may genuinely deserve most of the investment if its economics, demand and availability justify it. The problem is an account where nobody can explain why those products get the money while others do not. Concentration should be a decision, not an accident of the algorithm.
A product’s job can change.
Stock ages. A size run breaks. Returns come in higher than expected. A supplier delay moves the replenishment date. A season approaches its end. Any of these can change what the business needs from the same product.
A SKU that deserved growth investment last quarter may now be constrained by availability. Excess stock may make cash recovery more urgent than defending the previous margin. The advertising answer should change when the commercial facts change. Not because a dashboard colour changed.
The UKSoccerShop case study reports £520k gross cash recovered from ageing stock in a 45-day programme.
What to ask your agency.
You should not have to translate your business into a campaign acronym before the conversation can start. Three questions will tell you whether the account is being run around your products:
01Why does this product get budget?
Every meaningful allocation should have a commercial answer: the job the product is doing and the evidence that the spend serves it.
02What economics justify the spend?
Not a platform score. The margin, stock position, returns and demand picture that make this a sensible place for the next pound.
03What evidence would change the decision?
A recommendation without a stated tipping point is an opinion. You should know what would make the agency reverse it.
BOI® (Bid On Intent) is how we connect those judgements to the account. The jobs are decided with your team, not delegated to a platform. Read why we built our approach around Every SKU has a job®.
How it fits together
One principle. One method. Reviewed every week.
In fifteen seconds: decide what each product is for, advertise it accordingly, judge it on business outcomes, and keep checking whether that job still holds.
- 01
The principle
Every SKU has a job®
Each product should be advertised for the commercial job the business needs it to do, not for whatever Google finds easiest to sell.
Read more - 02
The method
BOI® (Bid On Intent)
How a product's job is translated into Google Ads decisions: structure, budget and the economics each product is allowed to run at.
Read more - 03
The vocabulary
The five jobs
Scale, Profit, Protect, Recovery and Gateway. A job is a human decision, reviewed as the business changes, never assigned automatically.
Read more - 04
The scorecard
Commercial measurement
Contribution, stock, cash and customer economics, judged alongside platform figures such as ROAS rather than replaced by them.
Read more - 05
The review rhythm
The Five Rounds
Our weekly review discipline: what changed in product economics, what decision followed, who owns it and how its effect will be assessed.
Read more
Can your agency explain why each product deserves its share of the budget?
Let’s look at the products, the economics and the decisions behind the spend. Not just the campaign score.
“But ROAS is green 👍”
If that is the only answer, we have more to talk about.
See what your ROAS is hiding