Delivery Optimisation vs Bid Strategy — The Difference Explained

Quick Answer

Delivery optimisation tells Meta what action to predict and bid for. Bid strategy tells Meta how aggressively to bid on that prediction. They live next to each other in Ads Manager but they control entirely different model components — optimisation drives the eAR calculation, bid strategy drives the bid value that gets multiplied against eAR.

The Mechanism Explained

Recall the Total Value formula: Bid × eAR + Quality. Optimisation and bid strategy each control one factor.

Delivery optimisation selects which event the eAR model predicts. Choose Purchase and the model loads the Purchase predictor. Choose Add to Cart and it loads a different predictor — same architecture, different training labels. The predictor outputs the probability that the impression results in your chosen event, and Meta multiplies that probability by your bid to compete in the auction.

Bid strategy controls how Meta sets the bid value when you don't enter a manual bid:

Lowest Cost (Highest Volume) — bid up to whatever's needed to spend the budget while keeping CPA as low as possible. No CPA ceiling.
Cost Cap — bid such that average CPA stays at or below your cap, sacrificing volume if needed.
Bid Cap — never bid above the cap, even if it means underspending. Hard ceiling per auction.
Highest Value (ROAS Goal) — bid to maximise total conversion value, optionally constrained by minimum ROAS.
Minimum ROAS — like Cost Cap but for return rather than cost.

The two interact. If you optimise for Purchase with Lowest Cost, Meta predicts purchase probability and bids whatever's needed to win. If you optimise for Purchase with Cost Cap $30, Meta predicts purchase probability and only bids when expected CPA ≤ $30, declining auctions where it can't hit the cap.

Practical Implication

For volume-hungry brands, Lowest Cost almost always wins because it lets Meta maintain auction participation. Cost Cap and Bid Cap are valuable when you have hard unit economics — but expect 30-50% lower delivery in exchange for predictable CPA. Do not use Bid Cap unless you've calculated your true breakeven manually, because hard caps shut delivery off entirely.

Real Numbers

Lowest Cost ad sets average 2.1x more spend than Cost Cap equivalents at the same daily budget
Cost Cap CPA stays within ±12% of cap in steady state
Bid Cap ad sets miss 18-40% of available auctions because the cap is below clearing price

FAQs

Q: Which bid strategy "trains" the algorithm better?
Lowest Cost — it doesn't restrict auction participation, so the model gets more data per dollar.

Q: Can I switch bid strategy mid-flight?
Yes, but it triggers learning phase reset.

Q: Does Cost Cap include all costs or just media?
Just media spend divided by conversions in the attribution window.

Q: What happens to Bid Cap during low auction density?
You'll see periods of zero delivery — the cap is too low for the available auctions.

Q: Does optimisation event affect cost or just delivery?
Both. Different events have different conversion rates, so eAR magnitudes differ, and your effective bid shifts.

Pix-Vu

Bid strategy choices only matter if your creative actually lifts eAR. If your product images aren't above category baseline, Cost Cap will starve and Lowest Cost will burn budget. Pix-Vu helps you generate creative that the eAR model rewards — clean, varied, conversion-ready product imagery at https://pix-vu.com.