What is Bid Optimization in Programmatic Advertising?
Bid optimization is the buy-side practice of setting and adjusting how much an advertiser offers for each ad impression, so that every dollar is spent on the impressions most likely to deliver the campaign goal at the lowest workable price. In programmatic advertising it covers the choice of bid strategy, manual or algorithmic bidding, bid shading in first-price auctions, and how a buyer responds to publisher floor prices.
How Bid Optimization Works in Programmatic
Every programmatic impression is sold in an auction that lasts milliseconds. A buyer's demand-side platform receives a bid request describing the impression (site or app, placement, device, location, audience signals) and has to decide two things: whether to bid at all, and how much.
Bid optimization is the logic behind that decision. It weighs what the impression is likely to be worth against the campaign goal, whether that is reach, viewability, a cost per action target or a return on ad spend target. It is the mirror image of yield optimization, which is the publisher working to get the highest price for the same impression.
Manual vs Algorithmic Bidding
There are two broad approaches.
- Manual (fixed) bidding: the buyer sets the bid, such as a fixed CPM for a line item, or a maximum cost per click. Google Ads describes manual CPC as a way to manage your maximum CPC bids yourself, with different bids for each ad group, keyword or placement. It gives control and predictable pricing, and it suits buyers who already know which inventory performs.
- Algorithmic (automated) bidding: the platform sets a different bid for each impression based on predicted performance. Google defines Smart Bidding as strategies that use Google AI to optimize for conversions or conversion value in every auction, with Target CPA, Target ROAS, Maximize conversions and Maximize conversion value as the core options. Display & Video 360 offers automated strategies that maximize a KPI while spending the full budget, or prioritize a target KPI over spend, with an optional average CPM ceiling.
Algorithmic bidding is only as good as the signal it optimizes toward. Point it at cheap clicks and it will find cheap clicks. Point it at conversions that are tracked badly and it will optimize toward noise.
Bid Shading in First-Price Auctions
Programmatic used to run mostly on second-price auctions, where the winner paid just above the runner-up's bid. Most exchanges have since moved to first-price auctions; Google Ad Manager completed its move to a unified first price auction in September 2019. In a first-price auction the winner pays what it bid, so bidding your true maximum means overpaying every time you win.
Bid shading is the response. The DSP estimates the price the impression is likely to clear at and lowers the bid toward it, trying to keep the win while cutting what is paid. Shade too little and you overpay. Shade too much and you lose impressions you wanted. Digiday has noted the transparency problem: buyers have to rely on whatever the vendor has said is the right price, because the shading model is not open to audit.
Floors and Bid Optimization
Publishers set a minimum price for their inventory. In the OpenRTB bid request this arrives as the bidfloor field, the minimum bid for this impression expressed in CPM. Floors can also be dynamic: Google Ad Manager's target CPM setting dynamically adjusts the floor price on matching inventory to maximize yield.
For the buyer, the floor is an input, not a target. A good bid strategy asks whether the impression is worth more than the floor to this campaign, and skips it if not. Negotiated deals are often the better route for premium inventory where open-auction floors run high.
When Bidding More Is the Wrong Answer
Bid optimization inside one platform can hide a bigger question: is this the right place to be competing at all? When every competitor bids on the same audience, clearing prices rise for everyone.
Criterion Global saw this with the grocery delivery app Gorillas in New York, where programmatic CPMs were inflated by a competitor "arms race" among rival delivery apps. Rather than outbid them, we moved budget into hyperlocal channels the competitors had neglected. Traditional out-of-home formats proved more efficient because competitors had not bid up their cost, and linear TV delivered conversions in a space the other apps had left alone.
That is the logic of the Criterion Global Budget Blueprint℠: decide where money should compete before deciding how hard to bid. Bid strategy then becomes one part of international media planning and buying, alongside programmatic advertising choices such as programmatic direct deals that take impressions out of the open auction entirely.