5 Questions: Get Ready for Digital Marketing and Personalization

This post originally appeared on the Hedgehog blog on January 31, 2017.

The first time a marketer sits down with their tech team after a new web platform is implemented, everyone typically looks around the table and says “now what?” Having a nice shiny new tool to use can be daunting, no matter how much experience the team has. Here at Hedgehog we’ve found the crucial first step is to get the obvious on paper (or file or whiteboard or Trello or Slack or post-it notes).

Our first step for any client is to answer 5 basic questions on digital strategy. These help provide the foundation for building out the personalization needs for Sitecore, a powerful marketing platform, although they work for other platforms that can be personalized like Drupal.

  1. How do you measure success, and what values of these measures do you need to achieve?

This may sound very obvious, but many companies have difficulty finding internal agreement on digital marketing objectives. In digital, website conversion often dominates the discussion which lends clarity to the charter of the marketing organization. But many other views may be just as correct.

Defining measures of success needs precision. Saying “site conversion” without defining what a conversion is, which visitors are measured (and which are not counted), how interim conversions are handled, or any of a half dozen other issues is not productive. And once agreement is reached, understanding the current levels of that measure and what the targets will be often leads to more rounds of discussion.

In the end you should have a simple document that anyone in your organization will understand, stating which measures matter, their definitions and their targets.

For example, your measure might be “anonymous visitor identification” where the definition is the percentage of non-identified visitors performing an identifying action such as signing up for the email newsletter, setting up a site profile, or registering for an event. This would filter out visitors whose identity is already known and allows for multiple endpoints.

But what about job seekers? Their traffic might skew your results. So you might define the audience as anonymous visitors who do not end up in the job seeker classification, which means your real analysis takes some time to figure out, since job seekers might take a few visits before they are classified properly.

Refining this down to a number where everyone is laser-focused on improvement frees up the organization from argument and conflict. This allows for better and faster decision making, which is the crucial factor in iterative improvement and a critical step in getting started with digital marketing.

  1. What information, if you had it, would best influence your decisions around customer engagement as measured by #1?

Marketing succeeds or fails through the quality of decision making. Marketers and systems make vast quantities of decisions; systematically improving them is what drives success as measured by #1. But many organizations do not have a clear understanding of what information would allow improvements in the decision making process. Would separating high potential website visitors from low potential make a difference? Or is knowing their gender more important? How about location? If you are not authorized to sell outside the US, knowing a visitor is non-US lets you shift marketing spend elsewhere and filter analytics data.

Ideally there is pre-existing analysis to draw from, but that is not always the case. The team focused on your success metrics needs to work together to develop the information needed for good decisions, whether this is customer profile information (gender, location), web metrics (source, path) or enterprise data (transaction history, predicted churn risk).

  1. What of #2 can I do today or in the very near future?

As great as it would be to live in an ideal world, we don’t. So next you need a reality check on the information you actually have and can use. Investigating what data is in production systems and can be utilized by your web platform will be sobering, but you need to confront reality here. It doesn’t make any sense to build your operations around anything other than what you can actually do.

This can be fertile ground for unexpected insight. Maybe when examining your CRM system for customer attributes you find someone has added a predicted value model that would streamline personalization rules. Or someone has added a few microservices that enable searches with up-to-the-minute specialist information across hospital locations.

A byproduct of this effort is your capabilities roadmap. Benchmarking digital marketing capability needs and defining the access to data and systems needed for better decisions can drive development, application and integration efforts for the foreseeable future.

  1. Given what I can do in #3, how should I define my customers to optimize decision making?

It might seem odd to wait on customer persona development until after a technical capability assessment is completed. But it is far easier to adapt to what’s available than to build out new capabilities. So while you wait for your development partner, you can leverage the technology already in place. We see this all the time with Sitecore, which many of our clients use, where personas and personalization can be quite powerful with just the data currently available. Sitecore’s profile cards, pattern cards, engagement values, and engagement plans can transform your digital marketing strategy long before you address the gaps you may have found (in the answers to question #3 vs #2).

Defining customer personas is both easier and much tougher than you would think. It’s easy to come up with ideas that make intuitive sense, and it’s easy to implement useful initial ideas that will have some impact. But it’s hard to define personas in a way that can be executed well and that enable good decision making. So hard, in fact, that we will have a separate post much longer than this one on the challenge of persona development.

A quick example helps illustrate the challenge. Let’s say you sell marketing software. How many different types of visitors will you have? Some are easy to list – researching prospects, technical validators, executive sponsors, consultants, interested stakeholders…the list of involved parties can be long. But others unrelated to a potential deal look a lot like the others – finance, students, marketers who will be users, trainers (who would need to roll the solution out), testers (who would be recruited from who knows where in the organization), or other employees from target organizations. Defining personas in a way that helps accomplish your goals as defined in #1 can be done quickly for some effect, but the full effect will need to tease out the small percentage of visitors that are extremely high value from very similar visitors that are not.

  1. What will I do differently based on customer definitions from #4?

Ultimately this is the foundation on which your day-to-day marketing operations will be built. Since your goal is to reach the objectives defined in #1, understanding how you will iterate content, testing mechanisms, design, navigation, search, social, imagery, or infrastructure to reach those goals will fill your days, nights, and weekends.

This list will change over time, but we strongly recommend putting your first 10 hypotheses down on paper so that everyone knows how your organization is planning on improving. This is also a big enough topic to warrant a separate post, but a few ideas include comparing long form and short form content, testing different hero feature message types (qualitative vs. quantitative, advantages vs. benefits, people images vs. product, etc.), placement of registration components or the number of items in forms, to name a few options.

But if you’ve followed the process down to this point, you should know the paths in the logical chain to make a good hypothesis: “I will improve (measure of success by some amount) by focusing on the (specific needs) of (customer persona), who will (respond differently in this way) through (changing this part of the experience).” If, for example, you’ve settled on location-based personas, the daily operations should reflect that. Lifestage-based, journey-based, role-based, attribute-based, value-based…each would suggest different actions to take to improve your measures of success.

Putting The Pieces Together

It helps to take a day or two away from the daily grind in order to put the plan together. A consultant or agency can help, both through moderating the effort and providing expertise from other clients. We at Hedgehog, of course, do this all the time. So if you feel stuck and need some help getting started with digital marketing, give us a shout and we can get you on the right path.

Looking At Website Personalization

I spent a fair amount of time on a blog post at Hedgehog on Sitecore personalization.  Its always an interesting exercise to deconstruct a technology system to understand how the parts work together.  Its even more interesting to try and explain it.  Years spent building and/or marketing applications help me visualize how everything was built, integrated, tweaked and generally made to work together.

In Sitecore’s case, you can see how a number of systems were built up over time and eventually made to work together.  First components were controlled by rule sets.  Then A/B testing of the component content was required.  Which meant a measurement system for deciding on the winner was needed, especially for those without transactional sites.  Then someone decided an anonymous visitor classification system made sense, which was added to the rule sets.  Next cross-visit automation made sense, which also fed into the rule sets.  Then of course email as an extension of the automation sounded good too.

As a system, though, it works very well.  While I’d like to see more a pure algorithmic approach to visitor classification, with some value calculations that are more advanced than the manual Engagement Value, the system as designed allows marketers to do a very effective job of sorting visitors into realistic groups and altering their experience.

Doctor Persona Matching

PersonasWe spent an interesting evening a few weeks back brainstorming on best-in-class elements of the web experience in healthcare. What would the modern hospital system do differently? How would services delivery evolve? How would patients choose doctors? Why can’t every provider system have real-time appointment setting and management like OneMedical?

While a lot of interesting ideas came out of it, many of which we incorporated into our Hedgehog Capability Maturity Model, one in particular really resonated.

When we choose doctors, we are given a little information. A name, a photo, maybe a biography, their credentials, and maybe some research papers published. A few might have a video. How in the world do we know we’re making a good choice?

The more we dug into physician finder experiences, the more it seemed this is an area with great room for improvement. One idea seems very easy to implement that we’re looking to try with one or two of our clients is the persona matching tool for doctor selection.

Wouldn’t it make sense to look for a doctor whose interaction style meshed with yours? Someone who made jokes to keep you loose, or someone very clinical and professional to make you feel more confident in your choices? As you think about this decision, where major changes in your health will rest in this person’s hands, it seems very odd that we do not have better tools for understanding how well a doctor might fit with our personality.

Right now the main mechanism to do this seems to be trial and error. And that is not a very good approach.

We see this persona tool following one of many different paths, but the common thread is a way to characterize patients and doctors in complementary ways, and help them match up with each other to improve their ability to communicate and interact. Sounds an awful lot like a dating site, but shouldn’t that be the model?

I haven’t seen this in the marketplace, but I don’t think it’s a long step to introduce it into the process. Maybe Match will license their algorithms.

“Personas” Image by Nata Branttes used under CC BY-NC 2.0

Fun With Twitter

TwitterLogo_#55aceeSo I have to say it’s been entertaining to spend real time learning Twitter. While I’ve used it for information consumption and keeping track of certain non-FB friends, I haven’t pursued its more conversational and collaborative aspects.

And it turns out you have to be pretty obsessive to make those pay off. When there’s a lot going on, it’s hard to put in the time properly, or to monitor goings on all the time. I’ve found myself falling back into broadcast mode using Buffer – a much easier prospect. In this model I use the time I would curating content anyway and apply it for outbound information.

So far, interesting. But the real connections aren’t yet there. My fault, since I’m not investing anything in them. But that will be the next stage.

Avoiding The Weakest Link

Hudsons Bay Map circa 1917
Hudson Bay Map circa 1917

If you don’t already read RetailWire, you should go sign up now.

One thing they really do well is ask insightful questions drawn from the day’s news.  A recent topic was Janet Schalk’s move from Kohl’s to Hudson’s Bay.  It really got me thinking about how important technology decisions are to the long-term competitive advantage of a retailer.

And that got me thinking as to why this was true.  In my view, the current capabilities of a retailer are the result of thousands of decisions big and small made over the years.  Some of the decisions made long ago constrain today’s operations.

Going forward, successful technologists will avoid creating this problem by choosing much more portable options across the board.  This could be through deployment options (clouds now scale easily and avoid the lock-in from buying hardware), payment options (subscriptions with short cancel clauses vs. enterprise software or long-term subscriptions), and better user experience (much shorter learning curves and smaller switching costs).

The first two are now very widespread – it looks like security concerns are the only real barriers at this point.

I’d say the third is where many battles will be fought.  Its so easy to create applications these days that functionality will become a commodity.  Great user experience driving short learning curves and nominal training requirements will become a huge point of differentiation.

By being portable, avoiding lock-in, and being easy to use, we’ll avoid becoming the pain point for CIOs and drive faster and more widespread adoption.

Image by Wyman Laliberte used under CC BY 2.0

Hello (again) world!


So it looks like ignoring my site for a while has caused fatal errors, including incompatibility between my backups and current WordPress capabilities.  So time to start over.

The timing is good, given I’ve started a new gig as CMO and co-founder of associate.io, a platform for mobile workforce management.  More to come.