No project plan includes a project management line item for “failure”. Yet with marketing analytics projects, there are some common areas that frequently lead to failure. Knowing where those trouble spots are ahead of time, and what to do about them, puts you in the driver’s seat for project success.
This month, we discuss technology and data risks to marketing analytics project success. Last month’s focus was on the human factors that can derail a project. Next month we look at organizational roadblocks. Always with how-to practical advice for overcoming these weak links in your Marketing Analytic project implementation.
Analytics by definition are rooted in technology and data. Problems in either area can result in missed deadlines and confusing, even useless, results. Following are some of those issues and the all-important how to solve them.
1.UNFAMILIAR, DISORGANIZED DATA
The first time that your analysts begin to work with new data sources, and attempt to stitch them together can be a frustrating experience. Not only will your team be unsure of the definitions of different data elements, but often the data will be difficult to combine. Substantial investigation will be required to determine which fields include what data. In addition, data linking approaches must be identified and tested.
When marketers rush the analysts at this point, mistakes happen. Substantial analytic work in the wrong direction is another mistake-filled area.
At the start of your project, you can shine a light on how the group plans to deal with unfamiliar data. Just by talking about it can eliminate a whole host of issues. A data dictionary is another tool to help assure a common understanding of the sometimes-misunderstood data language. Involving marketers in the data validation phase of the project also helps assure success. The numbers simply have to make sense. Once you are sure that the data adds up and stands for what you think it stands for (e.g. tot_rev_gross = total sales dollars before discounts), then you can move the project ahead.
The next project-failure-insurance step is to make sure that the analytics team understands how the project results will be used. Particularly, the analytics team needs to be aware of the level of accuracy that the marketers will need in order for their intended action to be successful. For example, the first email program that you send to a new group of customers may not require that every customer has a valid email address; rather, if 50 or 60% of customers have valid email addresses, the resulting information may be very useful to a marketing program. Communication is the key to making sure that the analytics team isn’t operating at a higher level of precision than marketers require to be successful with their marketing efforts.
2. CONCERNS ABOUT DATA QUALITY
Purists on your team are sure to complain about data quality – even refuse to work on data that isn’t up to certain standards. “How can we be sure the data is clean?” they often ask. Of course, any manually entered data has its share of issues and deserves some level of correction before any analysis takes place. But taking the time and effort to find and fix every error is not in the best interest of your Marketing Analytics project.
Other data quality issues arise when distinct and different data sets are merged, which by definition happens in nearly every Marketing Analytics project.
While the goal of a Marketing Analytics project is not to correct every instance of incorrect data, it can be very valuable to identify which data is incorrect and which data we believe to be accurate. If the analytics team can identify which data appears to be incorrect, that data can be exported from the analysis quickly, permitting the analytics team to move ahead and the marketing group to get the information they need in a timely manner.
3. PURSUIT OF THE PERFECT STATISICAL MODEL
Statisticians can get very excited about refining models in order to get a more precise answer. The risk is that the additional level of precision will not be valuable and the time necessary to refine the model will unnecessarily delay the project.
Make sure that the marketing team is involved in reviewing the initial statistical model and understands the strengths and weaknesses of the existing work compared to additional refinement. If a model predicts improved results of 40% compared to a random customer group, those results may be more than sufficient to deliver a winning marketing program. The difference between a model that predicts a lift of 40% compared to a model which depicts a lift of 48% may be irrelevant to the definition of your marketing project’s success. Only when marketers dig in and understand the results of the initial model can the marketing team make sure that the statistics do not get out of control.
4. LACK OF VISUALIZATION TOOLS
When all of the data manipulation is complete and it’s time to communicate the results and learning, the format used is often a never-ending set of Excel workbooks. When you have dug deeply into the data, it can be very exciting to you to show all the details and interconnections to your audience. However, for your audience, such a level of detail may be mind-numbing. Some marketers are highly quantitative and will enjoy rolling up their sleeves and getting into the data at the detailed level, while other marketers (as well as salespeople) need to be shown the bigger picture in a simple way first, before they ask for a limited set of details.
Overwhelm them at the start with excessive detail across pages and pages of numbers, and you risk “losing them at hello.”
The key to success will be to understand what the critical pieces of information are for your audience, and then displaying that information in the most compelling way possible. When done well, visualization and infographics can become a living legacy for the work you have done – shared across the organization and from office to office.
In the past three years, data visualization tools have become more commonplace. There are more, simplified and open source tools than ever before – tools that turn data into graphically beautiful charts, graphs and pictures.
The key will be to first understand the information that your audience needs, then develop some rough ideas on how to display that information in a compelling manner. Only then can you review data visualization tools available in the marketplace today and choose the appropriate tool to fashion a story that no one will forget.
On your next Marketing Analytic project, pay particular attention to the all-important technology and data components. Now, when you stumble upon one of these common roadblocks, you’ll have some strategies for steering a path to success.
Try some of these ideas and let me know how they worked for you.