Endorse Maximize your ROI, minimize your project risk Consulting
Is your Data Warehouse paying dividends yet?
To paraphrase Webster’s online dictionary, the word dividend is defined as “A share of profits received by a stockholder”. If we view your clients as the stockholders, then this would be a valid question to ask them about the data warehouse. Do you know if your marketing data warehouse is paying dividends? Assuming your data warehouse is not sending out quarterly checks, how would you otherwise know?
To determine whether your marketing data warehouse is paying dividends, you first want to ascertain if it at least is getting a good Return on Investment (ROI). So to measure your ROI, we want to look at more than whether the project finished on time and on budget.
Meeting your ROI
Operationally, if your data warehouse is collecting and disseminating data as designed, by default it is satisfying your ROI goals. But is the data warehouse really benefiting the business? If all the data warehouse does is churn out reports from the day before, your business users are looking at historical data that they can use to validate what events had just occurred. Don’t get me wrong on this. Looking at what took place prior is essential for understanding what the business landscape looks like. This is the what of something happened but not the why. Some decisions or assumptions can be made from this data but keep in mind, these events had already occurred and you can not change them.
Operational Data Reporting illustration: You’re an online retailer selling a full line of household goods, luxury items, clothing, etc. You receive daily reports on goods sold and inventory. The following table illustrates the basic reporting data you would expect from an “out of the box” data warehouse. From this data you may receive products by category with counts, by store, etcetera.
Paying Dividends
Your dividends start to accrue once you are able to obtain actionable data which is the why aspect of a situation. Actionable data gives you the ability to anticipate an event that is about to happen putting you in a position to capitalize on an opportunity before it occurs. This is a very advantageous position to be in which you can leverage the knowledge to work in your favor versus working with operational data which can only tell you something had occurred.
Actionable Data Reporting illustration: In addition to your existing reports, you now receive reports on who purchases your products. You are able to target these customers with like products for cross selling. Continuing with our example and focusing on customer Jones, the following illustrates the ability to take action by suggesting similar products of interest.
Increasing your Dividends
We’ve spoken about the Operational and Actionable aspects of your data where you can be in the position to anticipate events so what more can you do? We can take things a step further by being able to influence events. Influencing events is different than anticipating events. Anticipating events is your best guess at what might occur based on your data. Influencing events is your attempt at trying to guide your customer down a path you desire.
Influence Data Reporting illustration: You now have the capability to analyze purchase history and profile high-end customers to the point where you can recommend new products of interest not purchased before by those customers. You are able to target these customers with products of interest for up selling. Again focusing on customer Jones, we have now have moved into the Influential level where products are related based on analysis of historical data. The logic below in this example is targeting Jones with a Travel suggestion based on past transactions of upscale items that imply affluence.
This is not as easy to accomplish as it sounds, but certainly is not an impossible task. A key component for this to happen of course is that the appropriate data is available. To determine what data is required we would have to go back to the requirements and analysis. Not necessarily to see if the source of data was mentioned, but to see if the foresight for paying dividends was built into the data warehouse in the first place. If the vision did not exist up front, it is unlikely that this level of payback could be recognized without a significant re-architecture effort.
Having the Foresight
As mentioned above, having the foresight can come from a couple of places. Either the client or the business analyst must have the vision of what the desired state for the data warehouse is. Or, from a good consultant that has performed this type of work before. Having someone that has been there before can make a tremendous difference between your success and failure. Once the stake is in the ground, hard work and some trial and error will get you closer to your end state. Don’t be surprised to find that getting to and maintaining your end state is usually an iterative process.
Regarding sources of data, we know this is a very important aspect for any application. When trying to determine what data you may need, first take a look at what is available and work from there. You do not want to omit a source of data based on a rash decision. If you are looking at the system from a long term perspective, a source of data may not have an immediate need, but it should be accounted for and built into the current plan with the future in mind. An example of this data may be customer returns. That data may not have much of an impact in the Operational or Actionable phase, but when it comes to the Influencing phase, you may want to only target your most profitable customers for up selling and use the returns data to help make that determination.
Do not forget to give consideration to external data sources either. These data sources can be appended to your data which contain attributes about your customer that you may not otherwise have access to. You can learn interesting facts about your customers like their hobbies, type of cars owned, real estate, credit, and the list seems to be endless. Another simple but valuable set of data to append is zip code associated data which you can use to determine Demographic and Metropolitan Statistical Areas (DMA and MSA), radius, and ethnic/financial statistics by zip code.
Probably the most important source of data you have once the system is built is your own feedback mechanism. As you create marketing programs based on knowledge from the data warehouse, you want to feed this data back into the data warehouse and adjust your processes as necessary based on the new data.
This type of data warehouse should prove to be an invaluable tool in support of your sales, marketing, advertising, inventory and online presence to name a few.
Reinvesting your Dividends
As your data warehouse begins paying dividends, a wise choice would be to reinvest your dividends to keep your investment growing. Add value by finding relevant sources of new data, creating new ways of looking at and evaluating the data, and continuous performance improvement. Your data warehouse is organic in a sense and it needs to be nurtured and cared for if it is to grow and be successful. Take care of it and it will take care of you!
Ken Pohl is the president and founder of Endorse Consulting which specializes in minimizing the risk to a project’s ROI by mitigating potential problems before they arise. His many years of developing and implementing data warehousing, business intelligence and CRM systems has provided a wealth of knowledge as proven techniques. These techniques translate into real world not from a book “best practices”.
