Proprietary methodology · Technical detail
How a product gets its job, and how that job reaches the auction.
The technical detail behind BOI® (Bid On Intent)
BOI® isn’t another bidding strategy
Smart Bidding decides how much a click is worth to Google’s prediction of a conversion. BOI® decides how much that conversion is worth to your business, then makes sure the account is structured, measured and targeted so the bidder buys accordingly. It sits above the bidding strategy, not next to it.
What Google knows
- Conversion probability
- The conversion value you send it
- Historical account performance
- Auction and competitive context
- Audience and query signals
What your business knows
- Margin by product
- Stock cover and sell-through
- Cash position and ageing inventory
- Strategic and own-brand products
- Customer value after the first order
- Promotional priorities and seasonality
What happens when the two are disconnected
- Discounted lines convert most easily, so they absorb budget while the products carrying the margin stay under-exposed.
- Stock that needs to move keeps ageing, because nothing in the account knows it needs to move.
- Spend rises on demand that would very likely have converted anyway, and reported returns stay flat while contribution drifts.
- Finance and the advertising report describe the same quarter in two different languages.
How BOI® connects them
Commercial data from your business decides the job each product is doing. The job decides where the product sits in the feed and the campaign structure, what target it is bought against, and how much budget it is allowed. Google still runs the auction. It simply runs it against your commercial priority rather than its own easiest path to a conversion.
Same product.
Different job.
Nothing about the product changed. Stock cover, demand and the cash position did, so the job the product is being asked to do changed with them.
01
Profit
- Margin 48%
- Stock cover 34 days
- Demand healthy
02: something changes
Under review
- Margin 48%
- Stock cover 142 days
- Demand slowing
03
Recovery
- Cash is now the priority
- Sell within the contribution floor
- Time box the window
One blended target, or four commercial jobs
Before
Every SKU
Target ROAS: 600%
- SKU A: Scale, 400%
- SKU B: Profit, 650%
- SKU C: Recovery, 300% within the contribution floor
- SKU D: Gateway, Judged on new-customer acquisition cost
Drag to compare
Where the judgement actually sits
Don't confuse a hypothesis with a fact.What we know
- Brand is 71% of PMax conversions.
What we think
- Reported growth is overstated.
What we need to test
- Brand exclusion and an incrementality read.
Illustrative account. The structure is the point: a fact, a hypothesis and a test are three different things.
What a job change looks like on the desk
Campaign report
Performance Max
Revenue £184,204
ROAS 6.8x
Stock report
SKU 91824
Stock cover: 167 days
Reason
- Stock cover 143 days
- Sell-through -31%
Spend action
£4,200 → Recovery campaign
Illustrative. Job changes are human decisions, never automatic.
A worked example
Illustrative figures, not a client account
Product A
- ROAS: 800%
- Margin: 18%
- Stock cover: 3 weeks
Product B
- ROAS: 550%
- Margin: 47%
- Stock cover: 19 weeks
On reported return alone, Product A wins and gets the budget. Look at what the business keeps, and each pound of revenue from Product B carries far more contribution. Product A also has three weeks of cover, so scaling it mostly buys a stockout; Product B has nineteen weeks sitting on the shelf, which is cash the business has already spent.
That does not automatically make Product A wrong to advertise. It may be a gateway into the range, or a product the brand needs visible. The point is that the decision is a commercial one, and the account should be built so that decision can be expressed, rather than left to whichever product converts most cheaply.
Why blended bidding fails ecommerce brands
Google Ads, by default, optimises against whatever conversion value you feed it. For most ecommerce accounts, that means gross revenue. The bidder treats a £100 sale as a £100 sale, whether it produced £60 of profit or £2.
The result is mathematically predictable. The algorithm scales spend on whichever SKUs convert most cheaply at the highest reported revenue. Those are almost always the lowest-margin SKUs in the catalogue: discount lines, free-shipping triggers, loss-leader hero products. Revenue goes up. Contribution margin goes down. The account looks healthier on the surface and worse underneath.
This is the core failure mode of every blended-ROAS Google Ads setup, and it is the reason most agency reports show 5x ROAS while the founder's bank balance shrinks.
Target ROAS 600% for every product? Really?BOI exists to fix this at the bidding layer, not at the reporting layer. Reporting on POAS instead of ROAS is useful. It tells you the truth. But the truth is only useful if it changes what the bidder does next. BOI changes what the bidder does.
The five commercial jobs
Every SKU in a JudeLuxe-managed account is assigned one job at a time. The job is not a guess. It is the output of a calculation that combines contribution margin, stock position and the SKU's role in cart economics. A SKU never holds two jobs at once.
Scale
Strong margin, healthy stock, and an acquisition or LTV story.
Push spend aggressively. Maximise impression share. Expand into adjacent intent. The constraint is the margin floor, not the budget.
Profit
Margin-rich SKUs where every incremental click should clear a defined POAS floor.
Bid to a hard POAS target. Cap CPCs at the margin ceiling. Do not chase volume past the point where contribution per unit compresses.
Protect
Competitive auction, rising CPCs, or a tight margin window the brand cannot afford to lose.
Defend current position. Hold impression share at a defined floor. Block the algorithm from scaling past the breakeven line.
Recovery
Ageing or overstocked lines where working capital is sitting in the warehouse. Low stock alone does not make a SKU Recovery.
Buy the cash back, not the ROAS. Spend to clear the stock inside a stated contribution floor and a time box, then the SKU returns to whichever job its economics now support.
Gateway
Lower-margin SKU that reliably pulls higher-LTV cohorts into the brand.
Bid against a forward-LTV value, not first-order revenue. Cap exposure so the gateway funds itself within a defined payback window.
Most accounts run 50-65% of SKUs in Profit or Protect, 15-25% in Scale, with the remainder split between Recovery and Gateway. The exact mix is a function of the catalogue and the cash position, not a target.
The two signals that put a SKU's job up for review
A commercial job is not a one-time setup. It changes as the business changes. Two real-world signals put a job up for review in the BOI framework. The change itself is a decision someone makes and records, not an automatic switch.
Stock position
When stock on a SKU drops below a viable threshold (typically defined as enough units to cover lead time on replenishment plus a buffer) spend on that SKU is constrained. There is no point bidding aggressively for sales you cannot ship. A backorder is a worse outcome than a missed click. Low stock is a limit on spend, not a reason to make the SKU Recovery. Recovery is for the opposite problem: ageing or over-ordered stock that needs converting into cash.
Cash impact
When a SKU starts tying up cash (slow inventory turn, supplier price increases compressing margin, returns rising past a tolerance) its job is reviewed against the new economics. A product that was a Scale last quarter can be a Profit this quarter and a Recovery next quarter. The bid changes because the underlying contribution profile changed, and because someone looked at it and decided.
Both signals are pulled from the client's commerce platform on a defined cadence: daily for stock, weekly for margin and cash metrics. Crossing a threshold raises the flag the same day. No one waits for a monthly review meeting to find out that budget is still going to a product that has been out of stock for ten days.
Why this isn't just better reporting
The most common pushback we get from prospects is: "Couldn't we do this with a custom column in Google Ads?" Or: "Couldn't we just use a value rules layer in PMax?"
You can report POAS in Google Ads. You can layer value rules. Both are useful and we use both where relevant. Neither solves the underlying problem on its own.
The problem is structural. Google's bidder is built to optimise against the conversion value you give it. If the conversion value is gross revenue, the bidder optimises against gross revenue. If the conversion value is dynamic contribution margin, the bidder optimises against contribution margin, but only if the data feeding it is accurate, up-to-date and segmented by SKU.
That last clause is where most setups fall apart. The cost data in Shopify is often wrong, missing fulfilment costs, ignoring returns. The PMax asset group structure obscures which SKUs are getting which budget. The lack of stock-awareness means the bidder will happily spend hard on products you cannot ship.
BOI is the operational discipline that makes the math work. It is the cost data hygiene, the SKU-level segmentation, the stock and cash signals, the campaign structure, and the weekly review and reassignment discipline, running as one connected system. It is not a dashboard. It is a way of running the account.
How it is implemented
The commercial thinking only matters if it reaches the auction. These are the layers it travels through. Scope on any given account is agreed with you at kickoff.
- Feed segmentation
- Product data is cleaned and enriched, then custom labels carry the commercial job through to Merchant Center. Labels organise the catalogue; they do not set a bid by themselves.
- Campaign architecture
- Products with different jobs are separated where the economics differ enough and there is enough signal to run them apart. Splitting for its own sake fragments data.
- Query and demand segmentation
- Brand, category, competitor and problem-led demand are treated as different purchases, because they are.
- Product prioritisation
- Budget and targets follow the job: scale, protect margin, defend position, recover cash from ageing stock, or acquire a customer worth having.
- Conversion values
- Where the data supports it, the value sent to Google reflects contribution rather than gross revenue, so value-based bidding optimises towards what the business keeps.
- Bidding and budget allocation
- Targets are set per job and reviewed against live margin, stock and cash signals rather than a single blended account target.
- Measurement
- Tracking is validated before anything is optimised. Performance Max is read through native channel performance, asset group detail and search-term reporting, subject to Google's disclosure thresholds.
- Incrementality
- Where budget and account size allow, tests are used to separate demand we created from demand we would have received anyway.
A fuller view of the execution layer sits on our technical approach page.
Questions
Technical BOI® questions
Next step
See what BOI® would do to your account.
A 30-minute commercial review maps contribution margin to a sample of your SKUs and shows how the job mix could shift. No prep needed.