Segmenting Email With Web Analytics

mediapost - TRADITIONALLY, WEB ANALYTICS OF ON-SITE behavior and email metrics have peacefully co-existed in separate silos. Once a message gets its audience to a site, many email marketers believe their work is done. Stefan Pollard, director of email marketing best practices, and Dan Miller, manager of professional services, both of Lyris, Inc. argue for a real marriage of the two disciplines. Lyris, a combination of several recent acquisitions, tries to bring some of these metrics together in its own Lyris HQ dashboard. Pollard and Miller walked us through how email clickthroughs need to be followed and understood after users get to the site.

Behavioral Insider: How are email and Web analytics different?

Stefan Pollard: Segmentation from an email standpoint is really around identifying what group of consumers a particular message should be delivered to, and then what action you want that group to take. The only action from an email is to click through to the Web site. But the other half of that is getting the right message to the right audience. Often the information available to an email marketer is very restricted to information the consumer gives at the point of acquisition. With the addition of Web analytics, you get more in terms of the behavior that a person is taking. You sent them an email, you put them in a target audience and sent them a piece of creative that directed them to take an action with your Web site. Now they are performing a series of behaviors. Not only is one of their behaviors to not go at all, but another behavior is to visit and to not take the action you want — but to look at other things within the Web site.

Dan Miller: So Web analytics historically is based on server performance, quantity of traffic, and basic numbers like that. Basically visitors are a series of clicks. Web analytics currently is really taking all this data and session-izing it, looking across all the sessions and visitors to determine which clicks go together to form a session. And then we can start drawing some conclusions from that. We really parallel email without usually crossing paths. In email terms, segmentation means people having in common similar attributes when they signed up. In Web terms, it can be exhibiting similar behaviors. Let’s look at customers who reach a certain point in a conversion process but didn’t actually purchase.

Pollard: The opportunity is people who have visited the Web site taking a behavior and you feeding that information into your email application to trigger an event-based message. Someone downloaded a white paper. You want to pass that information back and send a transactional response message, and might want to use it as part of a drip campaign. People who downloaded this are also interested in that. And maybe three days later you follow up with them based on whether they clicked on any of those categories people click on. And you may see they haven’t responded to them, so you can present them with a different set of articles.

BI: How many site and publisher are actually integrating Web analytics and email?

Miller: The two separate pieces are common. Tying them together, I found, is rare. Traditionally, the Web analytics side works with blinders, focusing on one or two metrics that are common for that marketing channel. Customers who manage Google AdWords PPC campaigns focus on CTRs, and once the visitor reaches the site they might hope to tie a conversion rate back to that and at least know what percentage bought something. But they really don’t pay attention to what happens in between.

Likewise, email marketers are used to email metrics — open rates and CTRs. They judge success by how many people click through. But if a lot of them immediately leave the Web site then it is a false indicator of success. We are advocating tying these two tools together. Between those stated preferences [email] and implied preferences through behavior [on-site] we can fine tune a more targeted message.

BI: Do particular kinds of metrics come naturally from this marriage? Like users not staying long?

Miller: We call it a “short visit,” someone who views one page and doesn’t click on any links. That would be the starting point, at least finding out whether the landing page they were directed to from the email was a consistent enough message and a compelling enough call to action that the visitor felt they wanted to invest in clicking deeper.

Another common practice in Web analytics world is to assemble what we refer to as a funnel report. Often the data will take a shape similar to a funnel, in that you are assembling a series of pages or groups of pages that you would like the visitor to proceed towards conversion, and analyzing how many of them actually make it there and at what point they fall off.

To apply segmentation — to only look at email visitors from particular campaigns — can help tie these applications together. We can go back to the concept of follow-up messaging, possibly targeting specific visitors based on the point where they fell away from the process. A classic example would be, I added an item to my shopping cart, shipping was calculated, and I stopped. You might infer that I objected to the shipping prices. So maybe you follow up with a free shipping message.

BI: What are the challenges for implementing integration in the organization?

Pollard: It becomes a different way of thinking about your market. Often you have a marketing department with five to 10 people each responsible for a small piece.

Miller: Not only do you have this siloed approach, but Web analytics historically has been looked at more as a technical application. So IT is running the tools. Even if the marketing people are consumers of the reports. they don’t have a real interactivity with the reports. They get numbers spit out with some pie charts. They don’t really use it to evaluate the individual campaigns — much less to potentially extract certain visitors based on their behaviors.

BI: With that in mind, who can best make use of the integrated data?

Pollard: Hopefully it starts more at the senior level, and you have someone who can at least wrap their head around the concept of a truly integrated campaign. It is not just gauging the success of the email on how many clicks it drove — but how the landing page converted, how we re-messaged people who didn’t do what we wanted them to do. I think the opportunity cost of not doing segmentation is starting to catch up with email marketers. Marketers who are still trying to focus on list size and one-message-fits-all strategies are having all sorts of problems. Those programs are not as successful as they were several years ago.

What’s the Best Frequency? Who Cares

mediapost.com - ONE QUESTION THAT EMAIL MARKETERS continue to ask all the time is “How often should I email my subscribers?”

If you are a publisher, this is still a reasonable question. But for most other businesses, it’s so last-century, so old-school, so Web-1.0, so… you get the picture.

The better, though more complicated, question might be: “What demographics, preferences and behaviors can I use to drive a continuous program that maximizes the lifetime value of my customers?”

In the old, direct mail world, you sent to a list until it didn’t make money anymore. With email, consumers tell you when you’ve gone postal on them by opting out or hitting the “This is Spam” button. But this paradigm is driven by sending regular, non-personalized “broadcast” emails.

If you can move to a system of emails fine-tuned to your customer relationships, you can likely email less often without sacrificing ROI, or even send more frequently and you won’t be penalized by ISPs. Most importantly, you’ll become even more valuable and relevant to your customers.

Email is the Swiss Army knife of marketing, giving you multiple tools to communicate with prospects and customers. I thought of 30 different kinds of email messaging that can be sent while I was listening to a session at the Email Evolution Conference recently: everything from welcomes to trigger-based messages to one-offs to confirmations to cross-sell/up-sell messages.

This vast array comprises “lifecycle marketing,” where the impetus for sending a message isn’t just the product or service you want to sell but a trigger, event, need or other factor of your customer’s, combined with your organization’s offerings and goals. Further, many of these messages can be automated (see David Baker’s MediaPost column on triggers http://blogs.mediapost.com/email_insider/?p=595), meaning you create the email, set the parameters, and let the technology take over. When you shift to this kind of customer-focused marketing, you turn the concept of frequency on its head.

How One Multichannel Marketer Missed the Boat

A few months ago, I bought a refrigerated wine cellar from a multichannel retailer whose catalogs I have been receiving for 15 years and emails for one or two years. This cataloger sells only wine-related items, from books to glassware to custom wine cellars. A wine unit like mine is one of the most expensive items it offers. Previously, I had made only a minor purchase. Now, I’ve vaulted myself into a high-value customer segment.

You wouldn’t know it from my inbox, though. After I purchased the cellar, I received a basic order confirmation along with an average nine emails a month, none of which acknowledged this significant purchase. At least one promoted the exact cellar unit I bought!

My purchase should have put me, and other high-end cellar buyers, into a lifecycle program. I was easily able to envision at least two dozen individual emails, all related to my purchase, persona and behavior, and all potentially able to drive more sales. Here are just seven types:

1. Order Follow Up/Customer Support: Besides the confirmation, send “It was shipped” and “Did you receive it/have any problems?” emails. Each could also include some upsell message for extended warranty, etc.

2. Cross-selling/up-selling: Next, a series of emails promoting related products such as Riedel glasses, decanters, wine inventory software, premium openers, books, or tasting kits.

3. Product replenishment: The filter should be replaced every 12 months. Remind me early and around my purchase anniversary to change it with a link to the filter page on the Web site.

4. Special programs/offers: Send me a birthday reminder and gift-certificate program for my wine-loving friends. How about a special VIP invitation or discount to your regional wine-tastings and magazine?

5. Refer a friend, receive a gift: Most wine drinkers don’t do it in private. Some of my friends spend more than I do, too.

6. Check up: “How are we doing?” surveys, reminders to update preferences, post comments on the product.

7. Behavior-based: Where I click on the Web site or in the regular emails should trigger messages, especially if I abandon a cart.

Yours truly is a motivated buyer. But the nine-a-month, one-size-fits-all approach is not prompting me to pull out my credit card again. In fact, on average, I open only one or two out of the nine. Send me an email promoting the inventory software, and I’d likely pull the trigger.

So, the next time you are waiting for a plane or riding the train home from work, list all the email messages your company could be sending to add value to your customer relationships. I guarantee you can come up with 15 different ones, or I’ll eat this column!

3 themes shaping behavioral targeting

imediaconnection - As marketers test BT waters, they will need to come to grips with a still-evolving landscape and integrate some of the key lessons of the past 10 years.

The tantalizing promise of behavioral targeting — delivering ads precisely to the right internet users at the right time based on their past web behavior — has been touted for nearly a decade. While advertisers have been relatively slow to adopt BT as a mainstream marketing method, there are indications that this technology is gaining traction and may even be reaching a tipping point. A recent eMarketer report projected that advertisers will spend $1 billion on behaviorally targeted ads in 2008 and upwards of $3.8 billion on them by 2011. As marketers, and brand marketers in particular, start to test these waters, they will need to quickly come to grips with a still-evolving landscape and integrate some of the key lessons of the past 10 years.

Ad networks and publishers that offer BT develop audience segments based on common content consumption patterns (say “Sports Enthusiasts” who consume a threshold level of sports pages in a certain time period), then use information collected on individuals’ web browsing to categorize users in those segments. BT providers then sell advertisers media buys that deliver ads directly to segments likely to be influenced by them.

This approach requires a shift in thinking for many marketers. BT is about buying relevant audiences, not relevant context. If a user’s consumption of auto content suggests that he is in the market for a car, behavioral targeting offers automotive advertisers a way to reach that user with a relevant message when that user is not consuming auto content.

Benefits
The benefits of BT can be significant, especially in categories like automotive, travel and pharmaceuticals in which certain predictable patterns of online content consumption strongly suggests immediate interest in buying something:

• Behavioral targeting can make acquisition marketing easy and powerful by generating “lists” of good prospects that can be mined efficiently and anonymously.
• Behavioral marketing can be used on its own or in conjunction with other forms of targeting based on factors like geography or demographics.
• Audience-based targeting enables advertisers to reach audiences of interest beyond limited, and often expensive, contextually relevant inventory, thereby increasing frequency and the potential for cost efficiency.

Shortcomings
While the benefits of behavioral targeting can be impressive, marketers must evaluate whether or not BT is a good fit. Limitations of BT include these:

• BT is really good at generating small lists of buy-now prospects, but not so good at achieving mass levels of reach or addressing and developing consumers who are “up the funnel” from the immediate point of purchase.
• The lack of standard segment definitions across BT providers puts the burden on advertisers to cobble together disparate groups in an attempt to replicate its real target, with limited ability to plan and view a cohesive whole.
• For the most part those offering behavioral targeting have yet to define an application for marketing “upstream” — in the areas of awareness building and preference shaping that account for the bulk of media spending by large, sophisticated brand marketers.
• The direct response metrics typically used to measure and manage behaviorally targeted campaign effectiveness may not be relevant to brand marketers.

Three themes for the future
As more and more marketers turn to the web to help them achieve their branding objectives, the following three themes are likely to shape the structure of the industry and new product innovation:

1. Transparency
When assessing the validity of any behavioral target classification, it is essential for the marketer to understand the assumptions used to define the target. All too often the assumptions used are not particularly transparent and are shrouded in “black box” mystery. This is particularly an issue as firms work to standardize segment definitions in an attempt to increase their potential reach. Is someone who has visited auto content once in 45 days really an “Auto Enthusiast”? Increasingly savvy buyers will demand more transparency around qualifying behavior.

2. Privacy Protection
A global concern often associated with behavioral targeting is whether a user’s privacy is compromised in the process. The simple answer is no. Behavioral targeting only tracks where an individual has been online; it does not identify the individual through any personal data such as name, address or email.

However, consumers increasingly are concerned that personally identifiable information is mingled with non-personally identifiable information, which they fear might be used to discriminate against them at some point in the future. Groups representing a small handful of privacy-concerned consumers have pressured the FTC to make all cookies opt-in rather than opt-out, a radical approach that would bring the entire internet advertising industry to its knees. Unless the industry quickly develops privacy protocols that are effective, the government is likely to step in with solutions of its own.

3. Predicting Behavior vs. Reacting to It
Rather than picking off individuals who have “tripped a trigger” indicating they are in the market for something and attempting to influence their decision at or close to the point of purchase, the next generation of targeting will focus on helping marketers “fill the funnel” with more of the right people before they register interest.

One such approach is psychographic targeting. Marketers have long known that psychographic characteristics correlate strongly with brand affinity and buyer behavior, but they have had no way of targeting them in media. Psychographic targeting would enable marketers in highly competitive categories, where psychographic traits such as spontaneity, pragmatism and assertiveness spell the difference between brand devotees and the merely indifferent, to effectively turn up the volume against their most important consumers.

Without question, it’s an exciting time to be in the media and advertising business. Today, the leading behavioral targeting firms are effective list generators for direct response advertisers. But as brand dollars follow mass audiences online, behavioral targeting is destined to become a more important tool for marketers looking to efficiently and effectively reach their target audiences. Those firms that are able to define relevant segments for brand advertisers with mass reach will succeed.

Why are our targeted weekly e-mails getting little response?

Ask the Expert - When a marketing program does not work, always go back to the basics of marketing 101: right target, right message, right timing, right channel. Despite advances in scenario-based Web design created to track customer online behavior and build profiles, or personas, for targeted marketing campaigns, your customers are still bombarded with tons of irrelevant messages—often the result of batch-and-blast campaigns sent by stove-piped e-mail applications and outsourced e-mail service providers—each day.

If you aren’t getting responses, it is quite likely that your customers view your correspondence as spam, rather than relevant or valuable information, regardless of whether or not they have opted to receive your e-mails. Relevance is key in e-mail.

Assuming your target audience, timing and channel are right, it’s time to step back and take a closer look at your message and its relevancy. What about your personalization capabilities? If you’re relying on customer profiles and Web behavior alone, it’s time to expand your approach and take into account all customer interaction: Have you met face to face at a trade show? Spoken on the phone? Engaged in preliminary sales discussions?

Often these data are stored in two different locations. Customer profiles are housed within your Web and e-mail system, and real-world customer interactions are stored in the sales force automation (SFA), or customer relationship management (CRM) system or a lead management database. What you need is an enterprise marketing platform that will link the two. Adding this technological infrastructure links online data accumulated through Web and e-mail tracking with offline data stored in SFA/CRM and enterprise resource planning (ERP) systems to automate the communication process.

There are a number of platforms out there, however the best solution will manage and gather data as well as support more personalized, tightly coordinated communications across multiples channels, not just e-mail. This technology should be able to handle sophisticated marketing automation work flows—especially those supporting highly personalized, large volume outbound communications, and should integrate with back-end transactional solutions.

At the same time, the technology should continuously and automatically monitor response rates and digital body language to provide you and your sales staff with up-to-the-minute details on campaign performance. With A/B testing and measurable results, you’ll be able to devise more targeted, impactful campaigns while integrating e-mail with new and existing communications channels to increase customer loyalty and drive sales—a far cry from the soulless batch-and-blast campaigns flooding your customers’ inboxes.