Key Elements of Complete Personalization
I have been writing about using “model-based” personas stemming from a 360-degree customer view for proper personalization for some time now. This time, I would like to cover basic elements other than data and analytics. Too often, even advanced data players have a hard time executing personalization due to shortcomings in other areas. If consumers do not get to see personalized messages in the end, what good is all that data and analytics work?
Last month, I introduced a personalization framework that separates outbound and inbound personalization. Then I divided the inbound part into two groups again, one for cases where the target’s identity is known, and the other where the identity remains unknown. Such division is necessary as we are all living with marketing divisions created based on marketing channels and it is nearly impossible to identify all targets (refer to “Personalization Framework”).
Now, let’s examine another set of checklists for complete personalization. When I say “complete,” I am counting both “reactionary” personalization that is popular in the tech community; and “planned” personalization based on past transaction, promotion and response history, as well as demographic data.
Regardless of channels or types of personalization in the mix, marketers would need to connect all of the following elements to get the job done right (i.e., target consumer actually gets to see customized content through their preferred channels).
- First, starting from top-left, we need technology that enables us to show different content to different targets. If it is about the website, the site must be modularized. If it is about email, we should be able to swap different content in and out easily. If it is about mobile apps, such content drivers should be built in. If it is about online chat, then customized scripts should be triggered at the right time. If it is about offline, well then, marketers must train their store employees to ingest information from terminals or hand-held devices and pamper customers accordingly.The bottom line is that we need some technology to drive customer engagement. But one should never treat this part as the end game. Too many marketers fell into that trap, considering the job done by setting some commercial personalization engine on an auto-pilot. That is the source of many “bad” personalization efforts: ones that are annoying, invasive, irrelevant and, ultimately, boring.
- Moving clockwise, at the risk of stating the obvious, marketers must have an ample amount of content to display. In the days when commercial use of digital images and CGI (Computer Generated Images) is widely available, creating a library of content should be a matter of commitment. But a great many marketers suffer from content shortage, or on the reverse side, content overload, necessitating a decent content management system. There is no personalization, even after procuring the latest personalization engine, if everyone gets to see the same old generic images or messages.
- Then of course there are data. I have been talking about this subject ad nauseam, so let me just reemphasize that data must be the primary driver for all customized messaging. Various types of data from disparate sources must be realigned to create a “customer-centric” view (or commonly known as a “360-degree view of customer”), as personalization should be about the person, not channel, division or products. Too many marketers get overwhelmed at this stage, and sheepishly resort back to the default setting of a commercial personalization engine with rudimentary segments based on some intuitive rules. That is a real shame in this age of abundant data.
- Speaking of abundant data, to drive a personalization engine in near real-time, all of these datasets must be “summarized” in forms of personas, segments or model scores. Each score is essentially a summary of hundreds of considered variables, and they are in the end just another set of “small” data feeding into personalization engines. In the age of Big Data, making data smaller and more digestible using modeling techniques is an essential activity, not an option. On top of that, such statistical work also improves targeting accuracy. Even the worst model outperforms rudimentary rule sets designed based on human intuition.
Now the question would be “Jeez, where do we start”? Unfortunately, the answer to that question is “It depends.” It depends on the state of available data, technology platform, content library, types of developed models and segments and, most importantly, commitment level of the marketing leaders in the company.
If all of these four elements are in semi-decent shape, connecting the dots among them is the key for success. I’ve seen organizations where proper personalization is not being done even after excellent data environment and personas are developed and managed, because of fictitious barriers between departments and lack of a common platform to exchange results of data work with drivers of technologies for personalization.
For such cases, I would recommend a stepwise approach to build conduits among the key elements:
- Procure and install a commercial personalization engine (i.e., start with outbound email or inbound Web personalization, depending on channel strategies).
- Test the engine with simple segments (not necessarily model-based personas yet), or raw “trigger” data.
- Concurrently with Step 2, conduct a data audit to see if data sources are properly aligned on a personal level.
- Develop a 360-degree customer view, if it's not ready.
- Consolidate data around the 360-degree view, and convert transaction and even- level data into “descriptors” of individuals (refer to “Chicken or the Egg? Data or Analytics,” “It’s All About Ranking” and “Beyond RFM Data” for further details).
- Create personas in the order of importance based on marketing goals, channel strategy and product promotion schedules (refer to “No One Is One-Dimensional”).
- Map personas (and/or segments or trigger data) to proper content (e.g., match a "Wine Enthusiast” persona with contents for “wine”).
- Test a personalization engine with personas and tagged contents (similar to Step 2, but with model-based personas or model scores).
- Expand the practice to all channels (i.e., outbound, inbound-PII-known, inbound-PII-unknown, as described in “Personalization Framework”)
While there are other routes, there really is no shortcut in all of this. There is no magic bullet in complete personalization. Even this phased approach suggested here may not work for everyone. Just this morning, my team worked out a different set of steps to cover the basic four elements, as we were dealing with a highly matured analytical operation. But even such a company didn’t have all of the dots connected to handle inbound personalization properly, so we all had to step back and build a roadmap first.
The key is connecting all of the dots in the end. The order of operations may vary greatly, as we will all inevitably encounter different types of shortcomings in different areas. The simplest way is to identify the lowest-hanging fruit, then fully test it and check it off of the list. Challenging parts must also be examined from the beginning, as some groundwork may have to be done concurrently (e.g., start with data hygiene and tagging processes, while the content library is being developed).
Proper personalization happens only when all of these four elements work harmoniously. All of those sub-par personalized messages that we see in our inboxes, on mobile apps or on websites are the results of technology-driven efforts. It is time to bring some real data into the mix. With an organizational commitment, it isn’t that difficult to put it all together, taking one step at a time.
Collaborations among disparate departments and divisions can be fruitful in a short period of time, if there is a thoughtfully built technology-content-data-analytics roadmap. Without it, arduous arguments and ineffective division of labor are pretty much guaranteed, collectively heading to nowhere fast.
If you get to lead such meetings, please feel free to bring the chart that I shared here. When key elements for personalization are commonly understood, it becomes so much easier to prioritize tasks and share assignments among distinct teams.
Stephen H. Yu is a world-class database marketer. He has a proven track record in comprehensive strategic planning and tactical execution, effectively bridging the gap between the marketing and technology world with a balanced view obtained from more than 30 years of experience in best practices of database marketing. Currently, Yu is principal and chief product officer at BuyerGenomics. Previously, Yu was the head of analytics and insights at eClerx, and VP, Data Strategy & Analytics at Infogroup. Prior to that, he was the founding CTO of I-Behavior Inc., which pioneered the use of SKU-level behavioral data. “As a long-time data player with plenty of battle experiences, I would like to share my thoughts and knowledge that I obtained from being a bridge person between the marketing world and the technology world. In the end, data and analytics are just tools for decision-makers; let’s think about what we should be (or shouldn’t be) doing with them first. And the tools must be wielded properly to meet the goals, so let me share some useful tricks in database design, data refinement process and analytics.” Reach him at firstname.lastname@example.org.