The majority of the marketing and executive decision makers are convinced that using data to inform the decision making process is not only worthwhile but necessary to operate in today’s marketplace. However, information about how to get started in this space is lacking. The purpose of this post is to point out some steps that are essential in collecting data and using said data to inform decisions.
- Track your marketing efforts.
Without a cohesive tracking strategy, data collected will be difficult if not impossible to use. Campaign tracking must be planned and applied consistently across channels. Things to include in the tracking documentation are campaign, target audience, cost, ad copy or a content identifier, geography, and associated conversion metrics. For digital media there are also things like medium, source, and keyword. The list of metrics can go on and on, track your marketing at the lowest level possible (for example, the individual level) and also track information about that level (for example, male or female).
Don’t forget about offline marketing, while this data is not tracked or tagged in the same way as digital marketing; it is important to include in your tracking strategy. Be sure to identify methods to separate the application of different campaigns, especially where they overlap in application time. And again, track these efforts at the lowest level possible, often times this is at the DMA and target audience level.
- Store the data.
Choose a data storage system that works best with your IT and marketing teams. This is not Excel or Powerpoint (however if you have historical data in these forms, pull it out and insert it into the storage system of your choice). Traditional database systems can provide decent performance and can be integrated with modern day business intelligence and visualization tools. Be aware that marketing can produce a very large amount of data. There are many big data solutions available such as Amazon’s Redshift or Google’s BigQuery that can crunch through all but the biggest marketer’s data. Outsourcing data storage is a common choice among marketers because they can offer speedy integration, immediate response to changes and technical expertise that may not be available through your in house IT department.
We recommend spending a good bit of time designing your data model and considering the best data types for storage. Marketing systems output data in a variety of ways; developing tools that can consume and transform that data can take considerable time, and that much more if they have to be rewritten to react to a poorly planned data warehouse. As an aside, a lot of thought should also be applied to the system that will be responsible for ingesting the data; having a human do this manually is both expensive and error prone.
- Decide on attribution methodologies.
Be aware that marketing systems often use the attribution method that shines the best light on their channel or creative. Understanding different attribution methodologies in house is important. Consistent attribution across channels is important to understanding their relative values. Last click is a poor choice for many reasons but, without some heavy statistical work, understanding what the true channel value is, is a bit of an art. Google Analytics offers some defaults and the ability to select your own. More can be found here. Defining your methodology up front can inform the tracking and data warehousing strategy; this is especially true when you’d like to look at attribution statistically rather than using an artful rule based approach.
- Test your marketing efforts.
This is an important one. If you as a marketing executive want to understand the ROI of different campaigns and channels, it is important to look for ways to test and integrate these tests into your campaign planning. Think about different target audiences, seasonality, and other variables that affect your conversion metrics. A good post on this is here. Varying spend, the inclusion of different channels, and/or creative across channels provides invaluable measurement space. This isn’t a new idea for marketing, think about control groups and A/B UX testing. These ideas should also be applied to omnichannel campaign planning.
- Set and work toward goals.
Omnichannel marketing is no different than any other area of life. Without defined goals it is very difficult to measure performance or maintain consistency in improvement efforts. Set goals, define tactics, and give the tactics time to work. Define a plan with predefined decision points and don’t change tactics until those decision points are hit. Understand the conversion cycle and be very wary of changing the plan without providing adequate time for conversions to take place. Measuring the effect of a series of campaigns or a specific channel or tactic within a campaign becomes extremely difficult when the tactic isn’t given time to work.
When goals change, be aware that the plan was for pre-existing goals. Don’t try to measure a campaign against a goal it wasn’t planned for and be aware that insight can be drawn from these campaigns.
In summary, like anything else, data driven omnichannel marketing takes planning and thought. These steps will get you moving in the right direction but be prepared to wait. Depending on your sales and marketing cycle it can take months or years to have a robust enough dataset to answer questions as simple as “what is my ROI on paid search?” Don’t hesitate to run through these steps and get started!