In a utopian world there would be one data source that would contain all the information that marketers and agencies would need to track the performance of their digital media. Unfortunately, the reality is far from that utopian idea. As many frustrated analysts and digital marketers know, tracking the performance of digital media can be very time consuming and mind numbing. Why is this the issue? Data collection and sharing has accelerated at an unprecedented pace, nearly everything is trackable for digital media, and analysts have powerful open source tools at their disposal to make sense of the data faster. In this post I will highlight four reasons for why measurement has become overly complex for digital media and offer some suggestions on how to improve measurement plans.
Aggregation vs Integration
Most of the time spent by analysts at agencies is around reporting. Most of these analysts have mastered the use of excel and powerpoint. Some might even know more advanced analytics related to testing and optimization, databases, and predictive analytics. Still they are limited by the disjointed data-sets that they work with.
Agencies and marketers focus way too much on aggregating their data so they can present the information rather than investing in resources that can help them integrate the data. A typical analyst might have to work with agency partners to gather the data in excel files, ad-server data, DMP data, data provided by research vendors, data provided by other agency partners across different Geo’s. All these data-sets only tell a part of that story and very few analysts can spot trends easily. Furthermore this issue is amplified when marketers look at various channels such as search, website analytics, and display advertising in silos.
The focus for marketers and agencies should be around creating a unified data source. For media that is served through digital channels (and some offline channels), everything should be tracked in a Data Management Platform or a proprietary solution that focuses on reporting and analysis as much as creating audience segments.
Industry Challenges
Even if an agency evangelizes integrating their data, there are way too many players in the digital advertising space that work in a silo. For example, Facebook and Twitter place a lot of constraints on placing external tracking of the media they serve. All reporting is done through their interface or and data extracted from their APIs cannot fully be joined with the information inside a DMP. This limits the ability to get an accurate picture of reach, frequency, and in building models that can help evaluate the performance of these channels.
Mobile advertising spend is also increasing very fast and this channel is also measured in a silo. Recent advances by Facebook Atlas in knowing the same user across mobile and web will definitely help mobile breakout of this silo.
Questionable business objectives
This challenge is the hardest to overcome as some agendas are pushed from the top-down. For marketers that are not judged on the ultimate outcome metric (sales, leads, more sales!) digital advertising could be a total waste of money. Earlier in my career I worked for a startup where we wanted to become very effective at collecting leads and converting them later through email. Our goal was to increase the lead conversion rate, a top of the funnel goal. I primarily utilized adwords search channel but I couldn’t get more volume to our website. I decided to try out Adwords similar audience feature to build a display campaign that would give us more traffic and in theory a similar audience that would sign up for our service at a similar rate. After running the campaign for two months I analyzed both cohorts all the way through conversion and saw that the similar audience group had a 20% lead conversion rate and an overall paid conversion rate of 0.2%, while the control group had a lead conversion of 18% and an overall conversion rate of 3.5%. This lesson really stuck with me for two obvious reasons. One was always measure the most significant conversion event for your business and seldom create a measurement plan that tracks top of the funnel goals. The second was more data integration. If I had access to understand the audience make-up of both of these groups I might have identified the issue and could have tweaked my campaign for targeting the right audience.
For some businesses measuring sales goals can be hard and they need to rely on top of the funnel goals for optimization purposes. For them I recommend having a talented in-house team that can validate these top of the funnel goals and integrate data-sets that provide the analytical horsepower.
At this point some agencies might make the case that digital advertising is important for shifting brand perception or attitudes, but few marketers or strategist know how shifting brand perception can increase/decrease sales. In my professional experience working with major brands, very few of them tied sales to shifts in attitudes and for those that did, almost all of them gave digital advertising the least credit (think 4p’s!). It is very easy to track impressions, clicks, visits, and maybe even top of the funnel conversions, but it is extremely difficult to tie shifts in attitudes to real sales. Even if the agency or the client has the expertise to build models that can tie attitudes to sales oftentimes these models are rarely actionable and almost always don’t influence any optimizations for on-going campaigns.
Explosion of data providers
Companies that provide viewability, fraud detection, and brand safety solutions are the new kids on the block for 2015. Many of them have created a fear and hysteria that half of the display ads are viewable. While this is true, marketers and agencies need to understand how they can use these new data-sets. Will they provide any actionable results? Right now that answer is not clear. While advertisers armed with this new data might have more negotiating power when planning media with publishers, the methods to track viewability and fraud are far from ideal. The current IAB standards state that an ad impression is viewable if it is in view within the browser for one second. Even if this standard is agreed upon by many advertisers, there are lots of different formats that are not supported, and the measurement protocols across browsers is not standardized. Even worse, many of these new vendors are not integrated with ad-servers or with the conversion touchpoints for an advertiser. This means data is merely aggregated at a publisher level without any knowledge of whether a viewable impression improves the chances of a conversion occurring for the marketer.
Fight for more data and transparency
Agencies and marketers that build agile data solutions which integrate these various data sets at an impression level are best positioned to deliver highly actionable analyses. Agencies need to re-think their hiring process and ensure that there are talented analysts who can work with massive amounts of integrated data and build predictive models, engineers that can help build, maintain, and improve these data solutions, strategists that are analytical and experienced to ask the right questions, and project managers that ensure that the changing requirements and technology help keep the data reliable. The feasibility to integrate various data sources needs to be the guiding principle for creating a measurement plan. CMOs should also adjust their expectations and re-think the agency-client relationship to one that involves data-sharing, trusting agencies to prescribe solutions for the strategy, and truly understanding the measurement plan and how it impacts their bottom line.