DV360 Tips & Tricks
DV360 continues to dominate the programmatic ecosystem as the largest Demand Side Platform (DSP) globally, despite the ongoing efforts of The Trade Desk (TTD) & Amazon to cut the gap. Whilst Google roll out new features & capablities to its enterprise solution, there are a couple of subtle things that all advertisers / traders should be aware of when running DV360 campaigns. This blog post will call some of these out, both from points which should be common knowledge to those that may unlock incremental opportunties!
CAMPAIGN BUILD AND INHERITANCE
DV360 follows a fairly trival hierarchy: Advertiser > Campaign > IO > Line Item > Creative, with the exception of YouTube (AdGroup / Ad) & some of the newer emerging channels.
Therefore it is important to understand where targeting parameters inherit vs not & at what level you would want to set them. Things like Brand Safety & Viewability may seem like they should sit on the highest level, which is largely true. However in certain types of buys, this can be a barrier to responding to bid requests so its important to be careful here.
The biggest call out & mistake I see a lot reside on the IO level. Any changes you make to targeting (net new) on an existing IO, does not impact existing Line Items. This has been a day 1 rule from the early days of DBM & at least Google do call it out a bit better now in the UI. But this is a key consideration.
For example, say you want to exclude a specific browser on the IO level. You can feel free to apply this to the existing IO but you should also apply this separately to all existing Line Items underneath. This may be easier therefore to do via an automated solution like the SDF bulksheet or the DV360 API.
REPORTING SEGMENTation by AUDIENCE
Audience targeting is one of the most popular targeting mechanisms advertisers use in DV360 campaigns. The ability to target your own 1st party data, Google’s audience data or 3rd party vendor data is something every advertiser has tried at some point in time.
But a key misconception is that you have to target audiences in order to get insights from them through reporting. DV360 has 2 capablities which are very under-utilised but are super interesting to use.
The first is the Audience Profile, which can be found under the Audience Analysis tab. This allows you to pick a seed audience (1st party based) & see how it corresponds to all the other audiences within your own 1st party data uploaded into DV360 as well as everything targetable in DV360. This is also compatiable with CM360 / GA4 & DV360 activity based audiences, so huge amount of opportunites to segment. I will note that there is a large degree of cookie led matching going on here, which over time will become less accurate. But this view is super easy to use. You can pull this via an offline audience composition report.
The second is in an offline standard report, which is purely the ability to add Audience List as a dimension into your performance reporting. What does that actually mean? It means that if you are running a Line Item with absolutely no audience targeting, you can still get reporting insight on what audiences the impressions delivered were grouped as. This therefore allows you to understand potential new audiences to target in brand new line items or alternatively exclude from the existing, as you have your performance data to work with also. Use the Audience List Type dimension to quickly filter between Google audiences vs 3rd party. This is a report that Amazon in particular have perfected within their DSP as a common optimisation technique, but in fact Google have had this since day 1 in DV360 also.
The powerful ORDER ID REPORT
If you are running a conversion led campaign, more than likely you are leveraging a Floodlight to act as the trigger to attribute a conversion to your DV360 campaigns. Ignoring the larger complexities of CM360 vs DV360 attribution, the most popular technique to understand more about your conversions is to pull a standard report with additional dimensions like domain, time of day or browser.
But what if you are not using CM360 as an adserver? And what if your KPI is not an actual Floodlight fire, but something within it as a custom variable? There is actually 1 report that allows you to go this level in DV360 natively, which is the Floodlight Order ID report.
Not many people realise this report exists or believe it is more for Pixel QA. But actually this is a super powerful report and let me tell you why:
You can see for any Floodlight that you’ve assigned to your DV360 Line Item, further details of your attributed conversion, including all custom variable passback
The Order ID dimension is effectively a string, which you would have to manually break out in Excel. This is the equivalent of the “Other Data” dimension in Google Data Transfer files.
You can add in a few additional dimensions like App/URL to get even more visibility on a custom variable attributed level
And perhaps the most interesting point of all, as this is a DV360 report, this has conversion modelling on it by default, which means if you are using CM360 also, this report will have a different view to a CM360 Floodlight report
For the most savvy, there is some interesting things you can do with this report, tying it back to a CM360 unattributed conversion report & getting a bit of transparency into the art of Google conversion modelling.
ADVANCED REPORING & ANALYTICS
This is reserved for the more technically led users of DV360, with the ability to leverage more of Google Cloud. DV360 reporting in the UI is useful & has some interesting areas (as proven already). But programmatic is so much more than that, to which if you really want to run successful programmatic, you shouldn’t take things at face value & get under the hood.
For this, there are 2 main options for Google. There remains to be value from the Google Data Transfer files, despite the limitations put on them from GDPR / removal of User IDs. This log level DSP data of DV360 gives more transparency than the UI ever will, in particular for things like domain delivery. Ever been tired of seeing “Low Volume Inventory” in a domain report? This is not a thing in the log data & you even get Full URL to work with also. But the real pinnacle is merging this data with other datasets, whether that be SSP Logs or Ad Verification Logs. My recommendation if you are also using CM360 is to get the unified DT file which combines both worlds. The opportunities to analyse & plug this back in via API / automation is severely under-utilised by most users of DV360.
The other option is to use Ads Data Hub (ADH). This is Google’s Data Clean Room, which is undergoing some major changes currently. It has got a lot of flack over the past 12 months, largely due to its use case outside of Google’s owned & operated delivery. However for DV360’s sake, ADH was the original solution to log level data going away. Google have continued to invest more in this event level data through ADH vs the legacy file. A good example of this is Supply Path Optimisation (SPO), where there are much more options to get insight into DV360 through ADH. And you can curate audiences in ADH to push back into DV360 for use in campaigns.
The conclusion here is that big data still plays a huge role in programmatic DV360 buys, but requires a skillset / resource that may not be possible for all users of the platform.
YouTUBE
One of the theoretical benefits of DV360 is to run YouTube buys in the same platform as your regular non-YouTube activity (Display / Video etc). The ability to cross-frequency cap using forms of modeling is a big USP that any sales material will always point out.
There is always the question of paying the additional 3-4% platform fee to run YouTube via DV360 vs Google Ads for free, but one of the interesting areas for YouTube through DV360 does come in the form of campaign setup + reporting.
If you are running DV360 YouTube and using CM360 as an adserver, it is imperative that you traffic your ads correctly or else feel the wrath of inaccurate attribution. The option to link a 1x1 CM360 tracking ad to each YouTube Ad you create is the key here, so make use of thsi option in the setup.
Whilst YouTube does have its own attribution model with engaged view metric, as well as unique reporting within DV360 itself, this at least allows CM360 to track back to a specific placement that you have defined. This trafficking point is not just about YouTube, as every DV360 account has a CM360 backbone regardless of whether you see it or not.
And speaking of reporting, YouTube placement level reporting was something that historically was super manual whereby only the Video ID was exposed. Thankfully, Google have added the ability to pull channel name into reports now, which really should be looked at for optimisation sake, especially if you are running without any strong approach to brand safety.
On this note, recently VAC (Video Action Campaigns) have been under the microscope with the inability to opt out of non-YouTube delivery through Google Video Partners. For those who are wanting more visibility, use the YouTube All Placement Report to get Full URL / Mobile App reporting on your VAC activity, which is actually more transparent than regular display DV360 reporting. You can also add Inventory Source as a dimension to understand how much is served where. And similarly to Google Ads YouTube, if you want to avoid GVP in general, exclude googleadsense_without_youtube.com as a placement in your adgroups.
CLOSING THOUGHTS
Hopefully this was useful, DV360 is continously changing and with more functionality coming from the API, there will be a lot more options to work with in the future. But this doesn’t replace the day 1 nuances of DV360, which may differ to other DSPs but are important to keep in mind when running activity.