SSP

What an SSP does

A supply-side platform is the publisher's counterpart to the buyer's DSP: software that connects a publisher's ad inventory to many sources of demand and manages how that inventory gets priced, filtered, and sold. In the standard chain — advertiser → DSP → exchange or SSP → publisher — the SSP sits closest to the publisher, deciding which demand may compete for each impression and on what terms.

The working parts:

• Demand connections. Integrations with DSPs, exchanges, and agency desks, so each impression is exposed to as many qualified bidders as latency allows.
• Auction operation. Running the real-time bidding auction (or participating in a publisher-side auction via header bidding), applying floor prices, and returning the winning bid.
• Yield controls. Price floors by geography, format, and buyer; deal management for private marketplaces and programmatic guaranteed; rules for which advertiser categories are acceptable.
• Quality and safety. Filtering malvertising, blocking disallowed creative categories, and maintaining the publisher's ads.txt/sellers.json declarations that let buyers verify authorized sellers.
• Reporting. Bid landscapes, win rates, and revenue by demand partner — the publisher's window into who values their inventory.

Major SSPs include Magnite, PubMatic, OpenX, and Index Exchange; Google Ad Manager bundles SSP/exchange functions with the dominant publisher ad server, which is why it anchors most publisher stacks.

SSP versus ad exchange — a blurred line

Historically, an ad exchange was the neutral marketplace and the SSP was the publisher's tool that fed inventory into it. That separation has mostly collapsed: every major SSP operates its own exchange infrastructure, and most exchanges offer publisher-side tooling. The terms survive as emphasis — "SSP" stresses publisher services, "exchange" stresses the marketplace — but treating them as strictly different companies or boxes in a diagram misreads the modern supply chain. What still matters is the function: who runs the auction, who sets the rules, and what each hop charges.

That last question deserves attention. The sell side takes its cut between the buyer's bid and the publisher's net, and the same impression often reaches buyers through several resold paths. Supply-path studies have consistently found meaningful spread between what advertisers pay and what publishers receive. Publishers counter with transparency files (ads.txt, sellers.json) and by limiting reseller authorizations; buyers counter with supply-path optimization. Both sides converge on the same preference: fewer, more direct hops.

How publishers should run their SSP stack

1. Connect several SSPs, not one — and not twenty. Each partner brings partially distinct demand; beyond a handful, added partners mostly duplicate bids while adding latency, fees, and operational surface. Measure each partner's incremental net revenue.
2. Set floors from data, not hope. Floors anchored to bid-landscape data protect value on high-demand inventory; arbitrary high floors suppress fill. Revisit by geography and format — a floor tuned for US desktop strangles emerging-market mobile.
3. Keep authorization files clean. ads.txt and sellers.json entries should match reality exactly; stale or over-broad authorizations invite spoofing and depress buyer trust in your inventory.
4. Watch net revenue, not gross bids. SSP dashboards love gross numbers. The publisher's metric is net revenue per thousand impressions after all sell-side fees — by partner, by format, by geography.
5. Manage demand quality alongside yield. The highest bid from a malvertising-adjacent buyer is expensive revenue. Category blocks, creative scanning, and advertiser blocklists are part of yield management, not separate from it.
6. Balance auction pressure against user experience. More partners and aggressive refresh raise short-term eCPM and page weight together; the durable optimum includes latency, viewability, and reader retention.

Common mistakes

• Treating SSP and exchange as exact synonyms — or as strictly distinct. Both errors obscure the real questions: who runs the auction and what does each hop cost.
• Ignoring fees and supply-path duplication. Counting partners instead of net incremental revenue; authorizing every reseller that asks.
• Evaluating yield without user experience. A stack that wins another 5% of revenue while doubling ad latency loses the audience the revenue depends on.
• Floor mismanagement. One global floor, never revisited, either leaking value on premium inventory or suppressing fill everywhere else.
• Neglecting the transparency files. Broken ads.txt entries silently divert revenue to unauthorized sellers or cause buyers to skip your inventory entirely.

FAQ

Do small publishers need an SSP? Direct SSP relationships typically require traffic minimums. Below them, publishers reach SSP demand indirectly — through Google AdSense, Ad Manager, or monetization intermediaries that aggregate smaller sites. The website monetization path sequences this progression.

How is an SSP different from an ad server? The ad server (commonly Google Ad Manager) is the final decision layer: it takes direct-sold campaigns, SSP/exchange bids, and house ads, and picks what renders. The SSP is one source of demand feeding that decision. Header bidding exists precisely to let multiple SSPs compete before the ad server decides.

How many SSPs should I integrate? Test incrementally: add a partner, measure net revenue lift at constant latency, keep or cut. Most publishers find returns flatten after roughly four to six well-chosen partners; the right number is where incremental net revenue stops paying for incremental complexity.

What take rate do SSPs charge? Sell-side fees vary by partner, deal type, and negotiation, and are not always fully disclosed. Estimate each partner's effective take by comparing buyer-reported spend (where available) against your net — and prefer partners willing to state their fees plainly.

Why is my SSP revenue volatile? Demand shifts with advertiser budgets (seasonality, news cycles), DSP-side path decisions (a buyer consolidating away from one of your partners), floor interactions, and quality scores on your inventory. Segment by partner and date before assuming the stack is broken.