Segments Matter

We've talked a bit about segments previously, but I want to clearly underscore that segmentation should be your first view of any campaign effort. Thinking more deeply about who receives your communications can remake your entire approach.

For example, what if your business owners were only prepared to deliver a 15 percent discount to your entire customer base as a purchase incentive for fear of giving away too much top-line revenue? Yet you could show them a small subset of the audience that were high-value customers with significant spending patterns that would spend three times more this quarter if you offered a one-time 25 percent off coupon. Having the data and customer insight built into your segment models to deliver this kind of insight is where the most effective marketers are headed.

So let's talk about a few layers of segmentation you might begin with and a few you can build toward. Although you could slice and dice your customer base almost any way one could imagine, there are three primary ways to think about segmentation: demographic, behavioral, and fully scored. Although scoring is really a superset of demographic and behavioral, I'm going to take the three separately because there's a lot of art to great scoring models.

Demographic data is a great place to begin your segmentation thinking. Normally we can run a quick snapshot report to understand how our customer base is split by gender, family size, location, or email domain. These very interesting splits can be the basis for some solid content-sharpening exercises.

For example, understanding that your primary multipurchase buyer is most often a female with at least one child can help guide both the content and the offers you make via email. You might aggressively experiment with images that include families or offer replenishment-based ordering designed for busy moms. The details of how and what you test will obviously be specific to your business, but you see the point: isolate a specific demographic element, and then market based on that buyer's wants and needs.

This approach also allows you to send custom messages to specific domains in the event of any new ISP changes that might significantly affect your marketing effort to a certain set of customers. Yes, I'm referring to the Gmail users in your database. The never-ending march of enhancements within Gmail make it almost a certainty that you'll want to communicate exactly how your recipients should manage the next feature that impacts how, when, and where they receive your messages within Gmail. This practical take on segmentation allows us to improve our marketing game just as much as communicating effectively to moms.

Let's pause before we talk about behavioral segmentation to discuss the difference between implicit and explicit preferences as it directly relates to your demographic segmentation strategy. Explicit preferences are what your customer will tell you via a form or an interaction with sales. For example, when someone opts into your mailing list, they might self-report as a woman; clearly we'd have no reason to doubt their selection. Yet interestingly, there's an implicit application of gender that could prove to be a very valuable insight in creating your dynamic marketing programs.

I have one UK-based retailer who actively looks at explicit versus implicit preferences when deciding what items to highlight in its marketing emails. Even though Jane Smith might have indicated she was female on sign-up, if she continually views men's items—perhaps in an attempt to improve the wardrobe of a potentially disinterested significant other—then the retailer will migrate the content accordingly. Its results show a small lift in conversions, but this is more about subtly demonstrating to your customers that your marketing is paying attention at an individual level.

This type of implicit segmentation also puts us on the path to understanding how recipient behaviors are often the most powerful elements on which to track and act. The retailer in the preceding example looked very specifically at website page-view data—an excellent way to begin thinking about behavioral segmentation. It's the epitome of the old adage “Believe what I do, not what I say.”

Although the cart abandon and browse abandon emails we discussed at the beginning of this chapter are great for ecommerce retailers, there's an equivalent behavior-driven aspect to almost every industry segment and persona. For the B2B marketer, understanding the velocity of visits from multiple people in a single organization over a week or a month might be a significant indication that you're in the final consideration set for a request for proposal (RFP) or your outright selection. And maybe the outcome of capturing that behavior is twofold: a subtle, benefit-oriented email to the primary buying contact and a CRM call flag sent to an inside sales rep to reach out in person.

This focus on tightly segmenting your audience based on their actions is the underlying principle of behavioral marketing as a whole. Considering the full scope of what someone says, does, and buys can give a marketer critical insight, not just into that individual but also into others who might act and buy similarly.

And finally, let's dive into the topic of scoring and how that can overlay and deeply inform your segmentation strategies. Traditionally, the idea of scoring behaviors was a B2B tactic designed to bring some sanity to a long and winding buying cycle that might include multiple channels and dozens of interactions. The primary thinking is to score each key event with an appropriate number of points so we can build a long-term view of the process.

The number of points awarded might look something like the examples shown in Figures 10.1 and 10.2, from one of our customers.

You'll notice they've outlined scoring modifications on three key points: demographic, behavioral, and time. Although each is appropriately weighted based on how important it is to the end conversion, it's critical to recognize the entire process is also factored for time. This is because they clearly understood that if someone shows interest early but hasn't purchased in the first 90 days, then that individual almost never does, hence the –5000 points that are applied on the 91st day. This, in effect, removes that contact from the majority of their marketing programs almost instantly.

So each prospect at any moment during the sales cycle has a specific score, and all scores are ranked comparatively in groups as shown in Figure 10.3.

There's now an absolutely objective, numerical view of where every prospect is within the buying cycle. The final step in their analysis was to map together the scores and ranks, then calculate six months of conversion behavior together as a baseline. Although they looked at both overall scores (demographic + behavior), you can see in Figures 10.4 and 10.5 that they found the strongest correlation between the behavioral score and true conversion.

Armed with this data, they realized that their most important job was to close more sales among the top two segments of engaged users—A's and B's. They felt confident that if they sharpened their content and involved inside sales at the exact moment someone rose above a 750 score, then they'd drive significantly more revenue. And they turned out to be exactly correct.

By deploying smarter closing tactics, they saw more than three times lift in the number of conversions among the top three segments of prospects. You'll also notice that they decided to split those top three segments into finer groups (see Figure 10.6, where A+, B+ and B– were added to the original A–F range) in order to drive even more specific offers and incentives at buyers close to conversion. For a company that had little idea where its prospects were in the overall buying process, this level of quantification had a major revenue impact on their business.

It's clear how concerted segmentation allows us to isolate and market to very specific customers on a one-to-one basis. When we invest time in understanding our customer's intent and where it lies in the overall process, we can roll out an entirely new set of tactics and messaging to support stronger conversion rates.

Logic, Queries, and Data, Oh My

Although this segmentation is obviously helpful, there's an underlying aspect of logic we should cover quickly in the context of campaign creation. It's one that manifests both during initial audience selection and at each step during an automated campaign.

We apply this logic in the Silverpop platform via a query engine. Before you get too scared about being a marketer in charge of queries (normally of the SQL [structured query language] variety, and the job of an IT person), the best SaaS platforms out there make this a very simple proposition. It's merely a matter of stacking up various demographic, behavioral, and scoring elements into a logic layer and deciding when it applies to the data.

For example, a query designed to build an initial segment might look something like Figure 10.7.

You can see the query factors for five separate layers here, which should, therefore, create a highly specific set of users. You'll notice across the top of the screenshot where you can add any other number of variables into the equation. So if you wanted a larger group to receive a specific communication, you might simply remove the social interaction layer, which should have the effect of doubling or tripling the number of people who will receive this specific message.

This logic layer, therefore, creates the first cut of prospects to receive an automated mailing. Just as importantly, you can apply this same exact method to any decision point of an automated program. So the net effect is an increasingly relevant audience-selection method based on increased activity by the prospect. Our marketing is literally getting smarter as it engages more deeply with our customers—and we're architecting a constantly developing stream of content to support the exact moment in each buyer's individual journey.

Being able to layer this logic over real-time users flowing in and out of the buying process across all your channels becomes a very important new way to think about one-to-one marketing. Not only can a marketer pull ad hoc reports to see what specific audience counts look like by numbers of elements in a query; a marketer can also report on specific behaviors that are key junctions and how those relate to end conversion.

Automating for Epic Scale

By now you're probably trying to figure out how to put all these program enhancements into action. No marketer has a chance of using these tactics to scale up their efforts unless you have a serious automation engine. You simply can't—and shouldn't—ever try this in a manual way by scrubbing consecutive Excel spreadsheets and loading up a new list every time you send a campaign.

To truly move to a behavioral marketing view, you need an automation engine that brings to life all the data and segmentation work you've done to get to this point. And again, this is a time to choose a platform that can manage the amount of contacts you have in your lists today—and what you expect in the next 12 to 18 months—without sweat.

Although all these rules and logic are easy for a marketer to build, the horsepower required to keep them running smoothly isn't inconsequential. If you want to score a behavior such as email opens and you have a database of 20 million, make sure to review your vendor's capability to handle 3–4 million scoring events in the course of 10 to 15 minutes following a full-list send. Don't find yourself in the unenviable position of asking a B2B tool designed for lists of 50,000 people to do the volume of work required for a list of 50 million users.

The other great benefit of automating your marketing effort is that you can simply do more with the same number of people. We have a customer who manages an absolutely huge marketing effort covering nine products, a Silverpop database of over 100 million, a bi-directionally synched Salesforce database of over 15 million and in excess of 500 million emails sent annually with a full-time staff of three—yes only three—people. Clearly they're very experienced and have very complicated cross-sell and up-sell campaigns across all those various products, but they've slowly created an epic set of automated programs over the last 18 to 24 months, and they pay strong dividends in revenue growth.

Great Creative Matters More Than Ever

It's easy and tempting to get caught up in all this new thinking around behavioral marketing. However, it's also important to remember that this new approach means that your creative execution matters more than even now. Are you willing to go to all this work in audience selection and data hygiene to then deliver the same tired email newsletter every two weeks with 10 links? I certainly hope not!

If anything, you should be thinking about ways to 10 or 100 times your creative executions to these even more compelling customer moments. You should absolutely be thinking in terms of dynamic templates that have data-driven elements based on demographic or behavioral customer data.

Before we go much deeper into creative, let's talk about A/B and multivariate testing. There are lots of schools of thought on if, when, and how much you should be testing email content. A most practical approach should be defined by two primary elements: your current performance and how many people you have working on email.

If you're a solo practitioner supporting six business units and you don't know what day it is, then biting off true multivariate testing that could produce 36 or 64 versions of individual creative for a single campaign would be an absolute nightmare. If, on the other hand, you have a campaign that you think is underperforming but you can't quite sort out why, then diving into subject line and other spot testing makes great sense. Maybe look at specific calls to action, button colors, or using different images in your message.

Even though most platforms let you test on-the-fly using automated methods, one element not to mess around with is the send-from address. Your users come to expect communications from a specific email address and name; this is a key element when they're scanning their inboxes, considering which messages to consume and which to delete. Most marketers I know think the send-from address is among the first three decision points for any email's consideration—especially in the mobile channel, where most of us have open rates above 60 percent. So stick to testing the contents and subject line of the message, and don't bite off more than you can reasonably chew.

Beyond testing, the question of responsive design is one most marketers have fielded in the last 12 to 18 months. How your content renders on the major mobile devices is a critical question—but one that's fraught with asterisks, what-ifs, and fringe cases.

A simple example is the difference in how iOS and Android handle message width on a mobile device. iOS will display the message in auto-width mode, which in effect makes the content as wide as the widest graphic element regardless of how small the text becomes as a result. Conversely, Android pins the message to the top-left corner; if the message requires a side scroll to see beyond the available pixels, then that's how it works. Neither is right or wrong, but it clearly affects how you should design your content. Anchoring a call-to-action button in the right-hand column would not be smart for Android users. You also wouldn't want to reuse a 700-pixel-wide header graphic from your website in an email for iOS because the font would be reduced to something around 6 pixels, and will almost certainly require the recipient to pinch and zoom—and it's highly doubtful they even undertake that much effort.

These days I'm much more likely to recommend our customers apply some radical simplification to their campaigns as opposed to trying to factor for every one of the hundreds of screen sizes now between phones, tablets, and desktops. (And don't forget, true responsive design for email requires a separate HTML body for each width variation, so you're not going to want to shove 10+ versions into a 250k email—even in this age of broadband.) Beyond simplifying your effort, the benefits of clean, one to two product emails with clear calls to action extend all the way to recipient engagement.

Even in nurture programs, designing to optimize for the next step in the process is an absolute best practice. I recently saw an example at a European conference of a six-part mailing that only sent the next message to openers of the previous message. By the time the campaign was complete, just over 25 percent of the people were still following the storytelling effort, and the open rate on message six was an astounding 92 percent!

So although most of the examples I've used in this chapter focused on email, remember that designing great campaigns is a job we need to master in all channels. Often, this means leaning on colleagues across our organization, including key functions like customer support and outbound sales. You'll want to share your top-line messaging and existing digital-program details with your peers as they build out call scripts and offer matrixes to give to their reps. Making sure you inform these other customer-facing channels of any major promotions and/or data collection efforts underway is a great way to maintain solid working relationships across your company.

The bottom line is your prospects and customers are being bombarded by dozens or hundreds of other emailers every single day. Fighting for your share of attention requires an almost magical combination of audience segmentation, timely automation, and stunning creative. When you can discover the ideal combination of events that drive conversion for you, get to work automating it instantly—and continue to dial it in over time.