Inventory-SyncedBiddingArchitecture
How we connect stock levels, product margins, and fulfilment constraints to Google Ads in near-real-time. The engineering behind POAS.
The Data Gap Between Commerce and Advertising
Most Google Ads accounts operate with a fundamental architectural flaw: the bidding system has no awareness of the commercial reality behind each product.
Google sees a conversion value of £120 and bids accordingly. It does not know that the product costs £72 to source, £8 to ship, carries a 22% return rate, and after VAT leaves a contribution margin of £4.80. It certainly does not know that the warehouse has 3 units left.
This is not a Google limitation. It is an integration architecture problem. The data exists in your ERP, your warehouse management system, and your accounting platform. It just never reaches the bidding algorithm.
We built a system to close that gap.
Three-Layer Feed Architecture
Standard feed management treats the product feed as a flat data export. Title, price, image, availability. That is Layer 1: the catalogue layer. Most agencies stop here.
Catalogue Layer
Product titles, descriptions, images, categories, GTINs. The standard Merchant Centre feed. Where most agencies stop.
Commercial Layer
SKU-level COGS, landed cost, category return rates, shipping cost bands, VAT treatment. The margin reality behind each product.
Operational Layer
Live stock levels, warehouse location, fulfilment capacity, replenishment lead times, stock velocity.
Layer 2 enriches every SKU with its commercial truth. This is where COGS data from the ERP meets category-level return rates from the OMS and shipping cost bands from the 3PL.
Layer 3 adds operational constraints. A product with 3 units left should not be bid on the same way as one with 3,000. A product shipping from a secondary warehouse with a 5-day lead time has different economics than one available for next-day dispatch.
When all three layers converge, the feed becomes a real-time commercial decision engine, not just a product catalogue.
Replacing Revenue with Profit as the Conversion Signal
The critical technical step is replacing Google's default conversion value (gross revenue) with actual contribution profit on every transaction.
// Standard implementation (what most accounts do)
conversion_value = order_total
// Result: Google optimises for revenue
// POAS implementation (what we build)
contribution_margin = order_total
- vat_amount
- sum(sku_cogs)
- shipping_cost
- (order_total × category_return_rate × avg_refund_pct)
- payment_processing_fee
conversion_value = contribution_margin
// Result: Google optimises for profit
This is not a post-hoc reporting adjustment. The profit signal is sent to Google at transaction time via server-side tracking, which means Smart Bidding learns to find profitable customers, not just high-spending ones.
The impact is structural. A £200 order with 8% margin sends a conversion value of £16. A £90 order with 45% margin sends £40.50. The algorithm learns that the smaller order is actually more valuable, and adjusts bids accordingly.
Over time, the entire campaign gravitates towards high-margin products and high-quality customers. Without a single manual bid adjustment.
Stock-Aware Bidding Logic
Standard Google Shopping feeds update once or twice daily. That creates a dangerous window: products sell out, but ads keep running. Every click on an out-of-stock product is pure waste.
Our architecture introduces stock-level signals at multiple decision points:
Feed-Level Suppression
Products below a configurable stock threshold are automatically excluded from the feed before they reach Merchant Centre. No click waste. No manual monitoring.
Velocity-Based Bid Modifiers
Products selling faster than replenishment can sustain get their bids reduced to extend stock life. Products with excess inventory get bid increases to accelerate sell-through.
Replenishment-Aware Scheduling
When a purchase order is confirmed and stock is en route, the system pre-positions bids so campaigns are ready to scale when stock lands, rather than waiting for a manual restart.
Warehouse-Specific Economics
For multi-warehouse operations, the margin calculation adjusts based on which warehouse fulfils the order, accounting for different shipping costs and delivery timelines.
The result is a bidding system that behaves commercially: it spends more when conditions are favourable and pulls back when they are not. Not based on platform signals, but on business signals.
The Data Pipeline: From ERP to Auction
The technical challenge is not conceptual. Everyone agrees that margin-based bidding sounds better. The challenge is building a reliable pipeline that connects disparate systems:
Each integration point introduces latency and error risk. A COGS figure that is 10% stale makes every bid 10% wrong. A stock sync that runs once daily creates up to 23 hours of potential waste.
The engineering challenge is maintaining data freshness, handling edge cases (partial stock, pre-orders, backorders, multi-currency), and building monitoring that catches pipeline failures before they reach the auction.
Why This Architecture Is a Competitive Moat
Most Google Ads agencies operate at Layer 1. They optimise titles, manage bids, and adjust budgets. That is necessary but insufficient.
The brands that consistently outperform operate at all three layers. Their bidding system knows which products make money, which ones are in stock, and which ones are worth fighting for in the auction today.
This is not something you buy off the shelf. Each implementation is bespoke to the client's tech stack, product catalogue, and commercial model. A fashion brand with 60% return rates on certain categories requires a fundamentally different architecture than a supplements brand with 3% returns and 70% repeat purchase rates.
That bespoke engineering is why this approach works. And why it is difficult to replicate.
Want to see how this applies to your stack?
We will assess your current feed architecture, identify margin data gaps, and show you where profit is leaking through the bidding system.
Frequently Asked Questions
What is inventory-synced bidding in Google Ads?
Inventory-synced bidding is an architecture where stock levels, margin data, and fulfilment constraints are integrated into Google Ads bidding decisions in near-real-time. Instead of bidding blind on products that may be out of stock or low-margin, the system adjusts bids based on commercial reality.
How does feed margin architecture improve Google Ads profitability?
Feed margin architecture enriches your product feed with SKU-level cost data (COGS, shipping, returns rates) so that conversion values sent to Google reflect actual profit, not gross revenue. This enables Smart Bidding to optimise for contribution margin rather than top-line revenue.
Why does Google Ads waste money on out-of-stock products?
Google Shopping feeds typically update every 24 hours. If a product sells out between feed refreshes, Google continues to bid on and show ads for that product. Every click becomes wasted spend, and customers who land on an out-of-stock page generate negative brand experiences.
What is a POAS-based conversion value in Google Ads?
A POAS-based conversion value replaces Google's default revenue-based conversion value with actual contribution profit. This means the bidding algorithm sees the margin each product generates after COGS, VAT, shipping, and returns, rather than the gross sale price.
Stay informed
Get our latest insights on Google Ads strategy delivered to your inbox. No fluff, no spam, just honest thinking.
Unsubscribe anytime. No spam, ever.