Combatting the new bidding algorithm in DV360

Taking a look at the current bidding system, the following example illustrates the final winning bid:

Working on a Swiss client, you are in a bidding war for a certain auction whether that be a single domain or a group of sites or a deal, whatever it may be, and your competitor is bidding CHF2.

If you set your bid at CHF10, you will essentially win the bid at CHF2.01.

This system worked really well as it ensured that campaign managers and clients were getting great results without having to over think their bidding strategies, thereby allowing the Trading Desks to spend time on other optimisations. Unfortunately this is all about to change.

The change that Google, along with other DSPs, is bringing in means that if there are fixed bids of CHF2 and CHF10 respectively, the winning bid will be CHF10 and not CHF2.01, so a much bigger impact on client budgets.

All is not lost however and NMPi Traders have identified strategies that can effectively protect our clients’ interests, and these are going to heavily influence the way Remarketing set-ups are constructed. The manual work on campaigns is about to get busier as Traders now need to use their skills to find the sweet spots in order to win at the relevant and lower bid.

Bids will now need to be hedged to find out where auctions are being won, and this can be done by setting up different line item strategies and having various bids ranging from low to high.

In order to know for sure that auctions are being won at the correct price traders will have to unfortunately run one at a time or else they will just find themselves winning at the highest price each time. Another alternative could be day parting but that will prove ineffective as bids can differ throughout the day.

Once the ideal bid has been figured out and the sweet spot has been identified, clients’ interests will be protected.

An alternative granular approach is to outlay the specific domains and test them singularly in order work out where the sweet bid spots are per site. This approach will again take more time to set up, but ultimately it is a fruitful exercise in order to reduce wastage in the long term.

NMPi Traders are passionate about finding solutions for their clients and working hard to make sure that they get the best service, so spending time developing strategies that ensure that clients get real ROI and are not being overcharged, is something that will ultimately reap rewards.

Once granular strategies have been set up, a more robust approach can be adopted. This is quite simple: start high and lower bids each day until the day the auction is lost, and then it’s straight-forward to rank the bid back up incrementally until the true winning bid is found and the bidding threshold is identified. Again, it’s manual work but not too difficult to accomplish.

NMPi is always looking for ways to improve the Trading Desk, and its Traders are constantly evolving in excellence, so while this might be a major change in the bidding structure, it’s not insurmountable as it simply requires a bit more work. The end benefit of the exercise will be well worth it, albeit quite possibly not fully understood or appreciated!

We’d love to hear what others are doing to counter this change.