Death by Data Dump
When I first started managing digital media at Ford over 10 years ago (which a recruiter told me is similar to dog years, so let’s call it 70 years ago), I was amazed at the level of detail we could track about our customers’ behaviors. Not only could we see how people were spending time on our website, we could also see where they were wandering around the internet after they left our domain. Although not cutting-edge tech these days, cookies created a new way of buying media (behavioral targeting) and gave marketers an unprecedented amount of accurate data about how their customers spent their time online.
Naturally, it wasn’t long before we marketers were getting all kinds of reports from our digital analytics teams. We began spending hours reviewing mountains of spreadsheets and graphs, portraying dozens of different data sets cut hundreds of different ways, all the while debating whether the 0.03% improvement in the click rate on a banner ad on the Yahoo! Finance page was worth sharing in our next 1-1. [Spoiler Alert: It wasn’t.]
So fast forward … um, 70 years, I guess, and some fundamental challenges have evolved with how many companies continue to handle digital media analytics. Specifically:
• We started keeping track of waaaaay too many metrics. Just because you can keep track of something doesn’t mean you should, but the ability to see all of this granular information was overwhelmingly addictive – and we clients fueled the fire. “But how did only the married men 35-45 driving pickups in metro Tulsa respond to the ad? More info! More is better!” So, as a result of all of this data;
• Overwhelmed analytics departments had no time to actually analyze anything. Very talented people (who might be otherwise providing really insightful comments) started spending hours just making sure the combined Excel spreadsheets from 4 different vendors didn’t have any mistakes. “Hey, looking at cell AF343, it looks like the response rate on the ‘Holiday Ad #6’ 728x90 was 438%. Is that right?" Which unfortunately has led to;
• Most digital media metrics reports have no clear actionable conclusions. I can’t tell you how many times I’ve looked at comments next to a graph – only to find that it’s just a verbose description of what the graph is already communicating. “Yes, I can see on the graph that the number is up 12% … so now what? Should we do more of that? Let it ride? Is it maxing out? My kingdom for a recommendation!”
And this is the saddest part of all. We have richer data than we’ve ever had – and almost no insights on what to do with what we’re seeing. Having lived through the evolution of this, I completely understand why things developed the way they did. And if I’m honest, we collectively (clients, agencies, vendors) all have a hand in where we’ve ended up – but no one seems terribly happy with the status quo. So, going into 2019, let’s make a resolution to try something above and beyond the traditional digital data dump.
Here are 7 things I’d propose to improve digital media campaign reporting next year. Some of these may seem elementary, but if I hadn’t seen these issues surface consistently over the last 10/70 years, they wouldn’t be on the list:
1. Clearly articulate the goal of every campaign prior to launch and measure accordingly – if it’s an upper funnel play to build awareness, don’t buy the campaign on cost per lead or optimize to transactions. Sales are critical – and every marketing effort may lead to immediate results – but not every message is designed to drive conversion, so don’t fixate on the bottom of the funnel if that wasn’t the intent.
2. Identify the key metrics that can influence business and jettison the rest ¬– what is really driving the business and what is important to your success? Depending on your measurement horizon, this may vary (e.g. if you are trying to drive sales, Cost per Acquisition [CPA] might be paramount, while if you have a robust measurement for the Lifetime Value of the Customer [LVC], you may accept a short-term loss of a new acquisition for the long-term gain). But build your analytics plan around what you value – and don’t get distracted with the dudes from Tulsa, mentioned above, (unless you are the Ford F-150 brand manager).
3. Look for trends, not just raw numbers – you may need at least 3 weeks of data (depending on the size of the campaign) as a base to start to identify patterns, but even after a week there may be some low hanging fruit (Key Purchasing Indicators [KPIs] with Partner A at $4 vs $300 with Partner B). Trends are where some interesting insights may begin to form.
4. Set benchmarks for every campaign – every campaign is unique, but if you don’t have any historical performance internally, every vendor/partner can give you some combined competitive industry average, or an anonymized BIC competitor you can use as a target.
5. A/B test everything – there are lots of ways to do this without a lot of effort: Site A vs. Site B. Image A vs. Image B. Headline A vs. Headline B. Website A vs. Mobile Site B. Build the campaign with this in mind and you’ll have a lot of optimization opportunities as you go.
6. Commit to adding at least one additional insight to every report – can you crunch the data every reporting period to find at least one solid recommendation on something that will drive the business forward? Find a way to take your analytics to the next level in 2019.
7. Subscribe to Avinash Kaushik’s The Marketing < > Analytics Intersect – Avinash is an absolute analytics wizard and tireless evangelist for actionable insights over “data pukes.” He is fluent in both analytics and marketing. This is a must-read for anyone that touches analytics!
You don’t have to commit to building a bespoke attribution stack to improve your digital media metrics – just start to make a few changes that can free up your analytics team to actually provide some insights and affect the business. Everyone will win.