PMPs bring us closer to the vision of programmatic media: bidding on the right impression at the right place and time. Unfortunately, performance has been inconsistent. It’s left some advertisers confused, asking, “Why doesn’t my PMP work?”
It’s easy to get discouraged and give up on PMPs. But advertisers need to be diligent about managing PMP deals to make them successful.
PMP set-up can be very easy. A publisher works with its SSP to set up a connection with the advertiser’s DSP, the SSP generates a DealID with the parameters of the buy, and the advertiser places the DealID into its DSP to start serving ads.
Parties on both sides tend to set it up and forget it until it’s time to review performance. But problems will likely arise with any PMP. For instance, the publisher may unknowingly block PMP activity. SSPs categorize advertisers, and the publisher might have chosen to block a category where the advertiser is listed. Or an unforeseen reserve buy might have entered a publisher waterfall with a higher priority and cause a competing PMP deal to under-deliver. If gone unchecked, these problems will hinder the performance of a PMP.
Advertisers need to constantly monitor PMPs from the onset and ensure activity is consistent over the duration of the flight. Advertisers and publishers need to begin any PMP deal by estimating the volume of expected impressions. Advertisers should compare this estimate to the impressions coming through on the DSP and monitor the bid requests being made by the SSP (the DSP can show this). Any major discrepancies should be immediately discussed with the DSP, the SSP, and the publisher’s team, which should include programmatic and ad ops team members. It’s also important to monitor impression volume by creative, since issues related to creative sizes often come up.
When impression volume stabilizes, advertisers can make optimizations. Recency, frequency, time of day, day of week, geo, and browsers are big levers. But the one with the highest potential for success and insight is audience data. First-party data from retargeting pools or on-boarded CRM files will likely perform best but with limited scale. Look-a-like or predictive audience targeting can be a scalable and efficient performer.
Advertisers should also consider “second-party data,” or a publisher’s first-party data. Most publishers use DMPs to house their first-party data, which can be used via a new DealID by an advertiser’s DSP. Cherry-picking third-party data, while typically inefficient in open exchange buys, can actually also work well with PMP deals. But it is important to understand how each provider is getting its data and the costs associated it. Top performing audiences can be shared with clients and used to inform other media tactics and creative development.
At the end of a flight, a PMP may still not perform well. Perhaps the CPM floor, the cost of the DSP, and the price of the audience data are too high and make the buy inefficient. Or maybe the site itself is not a good fit with driving the advertiser’s KPI. This shouldn’t condemn PMPs. Advertisers may need to revisit sites being used, negotiate lower CPM floors, or consider new audiences. PMPs are still one of the best ways to truly target the right impression at the right place and time. Advertisers just need to be diligent and find the right mix that fits their goals.