You’re struggling to measure the ROI of a marketing campaign. How can you ensure your metrics are accurate?

Being clear on your goals for each campaign is ofter overlooked. For example, a reach campaign on Meta is not going to drive the same outcome as a direct response search campaign on Google. Always think about what the campaign is there to do before assigning a goal. Very often people judge campaigns on the same goal (sales), which is a mistake if the campaign type or platform isn't setup to do this. Thinking about the medium in which people will see your ads, and their motives will help define this. For example, someone scrolling on TikTok is not searching to book a holiday, but might be interested in seeing a beautiful destination, then search for it later when they are ready to book!

I've seen many business owners make a mistake by coming up with marketing campaign ideas during meetings without clear objectives. It's important to understand that your goals should come first, not the campaign. Setting objectives is important because it helps you know where you're going and what you want to achieve. It keeps you focused and helps you make better decisions about what to do next. Without clear objectives, it's easy to get lost or waste time on things that don't really matter.

I'd agree with Kate here in that the choice of metrics has to make sense based on the KBRs for the Campaign. I typically work "backwards" and start with the platform(s) that make sense for the KBRs, then I look at what KPIs would answer the KBRs. For example, for Meta I might want to know "what's the % of reach by brand for a multi-brand campaign?" Then I'll select the KPIs that support that KBR (Reach by Brand). Ok that's really simple example but this part of the campaign should be sort of automatic (you choose the Metrics that not only are avaiable in the platform but you want to make sure that the questions posed by the KBRs are answered by those metrics). Then you can quickly show success to the CMO/CDO.

Reporting clearly is key but reporting is one of the most challenging aspects of marketing. Being able to get the campaign performance on the same dash viz as the analytics data (seeing media cost in same context of the campaign by platform and brand). What I found that works best is a CDP, GCP, and Domo/Tableau/PowerBI/Looker for the viz and GCP BQ to get the transforms in place so that everything (performance by platform, analytics, profile segments, etc) are available in the same dash. Very few CDPs or CRMs really make this easy so I'd recommend building this yourself using AI tools in GCP.

old-school attribution is misleading in my view so I've been focusing on building attribution models using BQML or Vertex (or even SparkML). Building your own attribution models may seem difficult but with AI tools this is much more approachable. The key is getting all the data points to support determining the attribution. This is done by combining campaign performance, analytics, CDP profile segments into a single dataset (map/reduce in Spark). Then using BQML or similar tool to determine attribution by building models that take into account all the inputs.