Sifting through Suppliers: Practical SPO Strategies for DSP buyers

WHAT IS SPO?

Transparency into the bidding process is difficult to come by in programmatic advertising. In a world that is increasingly moving to 1st price auctions, advertisers tend to know how much they “paid” (all-in) for an impression and the result of winning that auction (clicks, actions, completions, etc) but they don’t often get visibility into the fundamental question of what sell-side intermediaries were involved in the transaction and whether would it have been possible to buy that same impression for less through a different supply path. Finding this most efficient buying opportunity is the concept of Supply Path Optimization or “SPO". 

THE BACKGROUND

With the advent of header-bidding, more than one ad exchange can now field the same bid request, causing DSPs to see multiple bidding opportunities for the same available impression. This proliferation of requests is problematic because, of the 49% of every programmatic dollar that goes to tech intermediaries and not the publisher, SSPs average about 8% but have been observed to take a much larger range of 6 to 25%. In some cases, SSPs resell requests to one another, meaning there are two middle-vendors who compound sell-side fees. Add to all of this that the number of bid requests which need to be processed per second (QPS) is one of the largest operational costs of any DSP and you have a piece of technology that has the possibility of generating a tremendous amount of incremental cost to the advertiser, for only marginal benefit to the end-publisher.

WHAT ARE THE PROS AND CONS OF HEADER BIDDING (SELL SIDE & BUY SIDE)

From a publisher standpoint, fielding multiple bids for the same impression is theoretically a strong yield management tool: 

• It maximizes the number of exchanges who can field an auction for a given bid request and thus maximizes the number of buyers and buying platforms who have access
• It provides the opportunity to field more inbound bids, potentially leading to higher yield for a given impression
• It helps provide a backup in instances where bids get dropped somewhere within the programmatic ether and avoids unmonetized impressions
• It improves the likelihood of a user-match with the buy-side (thereby increasing the value of the impression) in instances where cookie-loss or lack of another type of identifier would prevent a given SSP from matching to a user in a DSP’s taxonomy

On the other hand, Header Bidding can cause a Bidding Headache for DSPs and advertisers, namely in that: 

• It means they have to field multiple requests for a single impression
• It means that an advertiser might accidentally bid against itself and drive up an auction that they could have won for less money had they only bid once
• It means that a buyer may accidentally buy an impression through a sub-optimal (less-efficient) supply path and pay a higher aggregate price to win the bid than they would’ve through a different auction

HOW DSPS COPE WITH INCREASED AND DUPLICATIVE IMPRESSION VOLUME

The solution to inbound impression bloat for most DSPs is to build buy-side tech that pre-qualifies available requests for all users of the platform and attempts to avoid too much duplication, where possible. This aspires to help clients avoid bidding against themselves or buying impressions against a less-efficient supply path but it also helps DSPs save on overall processing costs. However, there are two key issues with throttling supply access across a wide customer footprint:

1. Most notably, an efficient supply path for one advertiser may not be ideal for another, due (in part) to the tradeoff between win rates and price, as well as the contractual terms that an advertiser may have with a given SSP and the user ID match rate with each supplier.
2. It’s difficult to de-duplicate the same inbound header-bidding request in real-time and any attempt to do so aggressively means occasionally jettisoning desirable, unique bid requests.

TACTICAL RECOMMENDATIONS 

The good news is that every DSP user has the ability to prune their own supply path substantially, provided their platform gives them full, rich, win, bid, and auction logs. Here’s how to solve:

1. Review auction log time stamps by domain and placement ID to find the universe of SSPs providing impressions for a given domain or set of domains (or apps)
2. Pull win logs against those domains from the SSPs in question and match up any relevant performance data (Conversions, clicks, completions, <INSERT CUSTOM METRIC>, etc)
3. On a per-app or per-domain basis, find the SSP that is giving you the best CPM and/or effective cost per KPI at a placement-level
4. BLOCK the ones that aren’t performing well or consistently generate the highest price winning bids

The above steps should help bring down CPMs and vendor fees by reducing the overall share taken by unnecessary middle-vendors AND reducing the number of requests that a DSP has to process (the most sophisticated DSPs will also compensate you for helping them save money on your behalf). Also, if a buyer sees pockets of impression activity that are performant, they should ask their account rep to work with the platform’s supply team to open up more of that supply (if available). Most importantly, however, the cost reduction realized by effective supply-path optimization should increase your overall media effectiveness significantly over time.