7 Tips For Successful Programmatic Campaigns

1. Less is often more

The first crucial factor is the correct setup in the DV360. Experience has shown that setups with more than 15 LineItems within an insertion order do not perform so well. Especially when testing different targeting, overlapping settings are anything but beneficial for the algorithm to fully develop. A less complex structure and consolidation of the LineItems is recommended here.

2. Data, data, data….

Clean and, above all, sufficient data is the be-all and end-all of a successful campaign. One basis for this is, among other things, conversions. The algorithms can work efficiently if they are stored correctly. If you do not achieve sufficient data/conversions (Google recommends > 3/day or 100/month), there are so-called micro conversions as an alternative. These can be, for example, a 20 second session duration, shopping cart conversions or a subscription to the newsletter.

3.It’s the mix that counts – the inventory mix

There are many myths surrounding the right selection of programmatic inventory. Purely open auction or better yet deals? The answer, as so often, depends on the campaign strategy. Depending on the funnel level, a different focus when it comes to inventory makes sense – traditionally, open inventory is used for a performance campaign, while more deals are used for branding and upper funnel.

4. Good things take time – the right expectations

This is perhaps one of the most important tips of all. The algorithms take time. Depending on how quickly and how many conversions are achieved, the algorithm sometimes needs a few days or a few weeks to learn and develop its full potential. We recommend waiting to evaluate performance until approximately 100 conversions have been achieved in a period of 30 days (guideline value).

5. The thing with in-app environments

Algorithms cannot “learn” in-app environments because standard conversions do not exist here. Depending on the campaign goal, you should consider whether these environments are a necessary part of the campaign strategy or whether you should simply exclude them.

6. Frequent optimizations

Every time changes are made to the setup, the algorithms have to start their learning process again. These changes also affect the CPA, which increases briefly until the algorithms settle down again.
If possible, avoid making too many targeting or setting changes at the same time. In diesem Sinne empfehlen wir eine Optimierung nach dem Prinzip “One Change at a Time”.

7. A/B Testing

If you want to find out which bidding strategy (fix bid or algorithms) makes the most sense for your campaign, an experiment (A/B testing) is suitable. In an A/B experiment, 2 different insertion orders are compared with each other. For this purpose, the audience lists are split 50/50.  This procedure prevents a user from being targeted by both insertion orders and thus distorting the results.
For better comparability and to be able to make better statements during the evaluation, the selected insertion orders should have approximately the same requirements. Tip: Use the same creatives, media budgets, targeting and frequency cap settings.

We are happy to provide comprehensive advice on the subject of “programmatic campaigns”.
kontakt@e-dialog.group

Source of featured image: Photo by Martin Woortman on Unsplash