Personalization Engine Series
Personalize Your Tech Stack (Whatever It Is) to Improve Customer Value Management
Remember those old geometry proofs you hated in high school? Or maybe you liked them like I did. Regardless, a geometry proof is a pretty good metaphor for the conundrum that marketers are finding themselves in:
If you’re a marketer today, you must build and deepen your brand’s relationship with customers via personalized communications.
Here are the facts:
1. Customers, not brands, drive engagement now—they choose the frequency and they select the channels.
2. You’ve got more customer data than you’ve ever dreamed possible—and the data’s volume, speed, and complexity sometimes seem insurmountable.
3. You generally must work with the tech stack that you’ve been dealt.
That’s right about where your brain fizzles out. How can you take the facts you know above, and create the personalized communications that customers desire and deserve? The ones that build a measurable business value for your organization? What are the logical steps?
But here’s the thing: One fact in that proof isn’t needed. Personalizing your communications doesn’t depend on a tech stack. Not at all. No matter what tech stack you own—Salesforce, Microsoft, Adobe, etc.—a tech-agnostic framework is the critical piece to driving you to your personalization destination. Really.
At The Lacek Group, we call this framework a personalization engine. You can rev it up and get there fast via jet. Or you can take it slow via moped. Either way, on this trip to personalization, you’ll make six stops, in this order: collect data, curate data, decide, design, deliver, and measure. This journey allows you to use a framework enabled by technology (whatever it is, today or tomorrow) to deliver personalization. And doing so will make all the difference between great personalization and OK personalization.
So why is personalization so important?
Consumers today realize that brands have their data, and thus expect brands to know and understand them. If customers determine that a brand doesn’t care enough to address them personally or, worse yet, takes their business for granted, that brand might as well tell those customers good-bye. To grow, businesses must invest in their relationships with customers, satisfying them personally and creating value by doing so.
The road map described above allows everyone on your brand’s various teams—direct mail, email, social media, etc.—to incorporate a common approach to personalization. In other words, following a shared map means that each area doesn’t need to find its own way; instead, all team members understand and follow the map similarly. That ensures that your company pays customers the courtesy of truly personal and relevant engagement. And that, in turn, can lead to brand love for customers who trust in and value your brand.
Let’s consider this example of how not to do things: A customer buys a rug. A month later the company sends an email to that customer, asking her to review the rug—and perhaps even to post a photo of the rug to her favorite channel. But there’s a problem: The rug is on back order and won’t be delivered for another month. Not only does the customer wonder how the retailer doesn’t know that, she’s lost faith in the company. Perhaps, if she’s annoyed enough, she’ll even cancel her order. Rather than provide a great customer experience, deliver a relevant message, and deepen the relationship, the retailer has done the opposite—and is no longer trusted.
That’s what happens when an organization is siloed; each silo operates its own system. A personalization engine, on the other hand, synchronizes touch points, and enables everyone in an organization to use a common nomenclature to deploy—via direct mail, email, social media, etc.—personalized, relevant customer communications. Additionally, over time, the same repeatable process and ever-more sophisticated signals can be used to finetune and grow a personalization engine.
If your curiosity is piqued, I invite you to read our six-part Personalization Series. It breaks down the personalization engine into its fundamental parts so that you can deploy it inside your own organization using your own tech stack. And if you need help to do so, please don’t hesitate to reach out. The Lacek Group is ready whenever you are to help rev up your personalization engine.
Julie Bustos serves as senior vice president of Marketing Innovation for The Lacek Group.
Data Curation: Four Ways to Connect The Dots
Remember that old connect-the-dots game—the one a parent or friend likely taught you as a kid? You draw a bunch of dots in a pattern of squares and then take turns connecting one dot to another. If you make a box with your line, you get to place your initial inside the box and then draw another line. And so on. The person with the most boxes at the end wins.
That’s a fairly apt metaphor for how today’s savvy marketers transform data gathering into marketing. First, you gather data, capturing and collecting numerous customer data points from various channels, devices, parties, and open data sources. Second, you extract the important information (e.g., customer identifier, event time, event content, event amount, etc.) from preliminary data parsing work. And third, you systematically store that data.
However, once you’ve captured, cleansed, and stored the data, rarely can marketers directly apply it to their efforts. Instead, a data person must purposefully curate the information so a marketer can connect the dots, extract the meanings, and then take action.
In short, I’m talking about curating data using four distinct lenses to better understand customers and take action. The lenses can be managed independently, but they work together to provide a valuable view of the customer for at-scale personalization. Let’s take a deeper look.
1. Customer identification
To deliver personalized experiences and messaging at scale, a brand must identify customers across channels. And that can be difficult. Most people use multiple browsers across many devices and also engage offline. So, building a 360-degree profile of each customer requires stitching together data from various sources into a single view by identifying the same customer across channels (deterministically or probabilistically) via identity mapping. This process enables a brand to understand customer behavior across touch points and then deliver relevant messaging and experiences to that customer, no matter the browser or device, or even in person.
2. Customer preferences
To meet customer needs, an organization must leverage its knowledge of their interests via zero-party data—optional information customers explicitly share with a brand, e.g., preferred scents, favorite products, communication opt-ins, etc. An organization can also apply artificial intelligence/machine learning (AI/ML) modeling—such as content affinity, offer predilection, communications timing, the inclination of responding channels, and more; this modeling is based on behaviors of customers and look-alikes to implicitly project their preferences. Meeting customers’ expectations and needs hinges on understanding and leveraging their preferences.
3. Behavioral signals
An organization can see great results by interpreting behavioral signals to trigger timely, relevant, and purpose-driven communications. With e-commerce, strategies like website retargeting, cart- abandonment emails, and real-time product recommendations are key. Meanwhile, traditional CRM practices—such as lifecycle targeting, look-alike prospect acquisitioning, and behavioral signals—play an essential and irreplaceable role in serving highly targeted messages to segmented customers.
It’s important to point out:
When customers are in motion (customers initiate actions and signals indicating they are or are likely in the market), the timing of communication delivery is key. A trade-off can be made between simple creative designs with a focal message and elaborate creative designs with carefully preplanned content.
Just as customer preferences are explicit and implicit, so are behavioral signals. Explicit signals are direct observations of a customer’s present and past behaviors. Implicit signals are speculations of a customer’s future or near-future behaviors. In fact, AI/ML algorithms have made astounding progress and success in identifying and tapping into implicit signals. In addition, technology and analytic investment to contemplate a customer’s next move, plus wants and needs, will continue to be important.
4. Strategic segments
The first three lenses of data curation convey the customer perspective. Strategic segments, however, are viewed from an organization’s perspective. This is where a customer’s wants and needs intersect with a brand’s goals. To calibrate marketing efforts to the value of customers, an organization creates insightful customer segmentations based on their strategic value to the business. Strategic value can be defined by a customer’s value (current and potential future value), a customer's motivational fit (such as customer satisfaction, brand loyalty, lifestyle preferences, etc.), or product/service alignment with a customer’s wants and needs. Strategic segmentation differs from granular audience segments created by operations; the latter are aimed to provide micro-moment engagement with a set of small audience groups. Strategic segments provide the organization with the framework to measure and optimize its grand marketing performance, and from which to drive and deliver sustainable organizational growth.
In summary, a well-organized and centralized data-curation plan adds tremendous efficiency and consistency to personalization efforts. It paves the way for measurable, scalable, and reliable personalization across technology platforms, execution channels, and marketing stakeholders. In other words, setting in motion a plan to centralize data curation within your organization will help you build and, more importantly, activate a 360-degree view of your customers. And that’s a little like winning the connect-the-dots game over and over again—on an enterprise scale.
Shi Bu serves as senior vice president of the Data Intelligence Practice at The Lacek Group.
Making a decision can be a simple task, like what to eat for dinner, or a monumental choice, like finding a spouse.
Decisioning, on the other hand, is decision-making by means of computational methodology. These days decisioning is more high-tech than ever, perhaps especially in marketing. That said, what we’re trying to determine remains straightforward:
WHO GETS WHAT?
create micro segments to engage customers within the customer journey
determine and prioritize messages per customer
select personalized offers (including offer type and offer value)
WHEN AND HOW?
decide when and how often to communicate
choose the engagement channel(s)
set campaign orchestration rules within the engagement channel(s)
set campaign orchestration rules across channels
Additionally, a key part of the process is logging and tracking our decisions to enable performance-rule correlation analysis that will inform future decisions.
Before we dig deeper into decisioning, let’s take a quick look at data curation and its four lenses—customer identification, customer preference, behavioral signals, and strategic segments. They work together to support decisioning.
In the age of customer-driven marketing, it’s hard to imagine effective messaging that only takes static customer information into account. Today’s marketing practices should reflect that our customers can belong to a multitudinous and static group—yet be fluid, dynamic, and singular.
Strategic segments enable brands to create insightful customer segments based on the strategic value of the customers within them. Since strategic segments tend to be high-level and static, they need to be broken down into microsegments (sometimes called engagement segments) for scalable personalization efforts. This tactical view is based on behavioral signals, customer preference, user status, and situational/contextual characteristics, such as emotions, weather, location, etc. Personalization is based on the microsegments in which a customer resides at the moment.
How the four lenses of data curation work together to support decisioning:
As discussed in “Data Curation: Four Ways to Connect the Dots,” the previous chapter in this series, strategic segment planning focuses on the big picture: Which customers are most critical for future growth? How large an opportunity do they present? And how much can and should we invest? Strategic segmentation provides answers to those questions.
Meanwhile, tactical operations focus on how to acquire and engage individuals within strategic segments. Just because two individuals occupy the same strategic segment doesn’t mean they should be treated identically in terms of personalized engagement. We should create granular microsegments to humanize the customer journey. That’s how strategic segments become actionable and make a significant impact on customer engagement.
Given the volume, variety, and velocity of data, traditional if-then decisioning isn’t sufficient to meet today’s marketing needs. The conventional way to create orchestration rules (often referred to as contact strategies) has been to manually create a series of “if-then-else” rules. In other words, if x happens, then y happens. For example:
If it’s been five days since the initial communication and a customer hasn’t responded, then send a follow-up email or else do nothing.
If a customer belongs to this segment, then send these three messages, or else if the customer doesn’t belong to this segment, then send the default messages.
If an elite member has done this and that, then assign the member to segment ABC; or else if an elite member has done this but not that, then assign the member to segment DEF.
These strategies worked when decisions were based on limited data points and limited channels. But today’s ever-evolving, rapid-paced, Big Data marketing environment provides—and demands—much more information. We’re still tasked with sending “the right message to the right person at the right time,” but that’s become more challenging and complex.
What can we do to sustain decisioning? Craft a plan to leverage artificial intelligence/machine learning (AI/ML) to help scale decisioning.
Companies are already deeply invested in AI/ML. For example, Netflix leverages it to recommend content to subscribers. Persado uses algorithms to write and optimize email subject lines. And Salesforce’s Einstein engine can determine optimal email deployment times.
Indeed, more and more organizations are accessing or building decisioning engines to prioritize offers, deliver messages at the optimal time of day, select the best content for each person, and more.
As savvy marketers, we must embrace technology to liberate us from repetitive tasks. Then we can put more effort into complicated functions where the execution and evaluation of outcome still rely primarily on human judgment. The long-term success of decision automation depends on incorporating human feedback into AI/ML platforms.
Next, we must establish a feedback loop to evaluate and improve decisioning. How do we know our decisions lead to the desirable outcomes we expect? To evaluate the tangible correlations between decisions and outcomes, we need to log our decisions and then establish feedback loops to track and measure performance against them. The outcomes of the analysis can then be used as new data to refine future decisioning.
The need to automate and scale decisioning—and the solutions to do so—will continue to develop. Although AI/ML will play a growing role, human judgment will become increasingly crucial as a way to mitigate the bias introduced by AI/ML algorithms and marry nonlogical factors with logical actions.
The success of sustainable decisioning relies on the implementation of a blended and balanced decision-making model in which human feedback is incorporated into AI/ML platforms to determine the final action.
Wouldn’t it be great if we could apply the same precision to choices in our personal lives?
Shi Bu serves as senior vice president of the Data Intelligence Practice at The Lacek Group.
Connecting Design to Decisioning, Delivery and Measurement
You’ve found yourself at step four of our Personalization Engine Series. So far we’ve introduced the importance of a personalization engine. We’ve connected the dots, explaining how data curation via four distinct lenses (customer identification, customer preference, behavioral signals, and strategic segments) leads to understanding customers better and allows a brand to take actions toward delivering at-scale personalization. And we’ve emphasized the ability of artificial intelligence/machine learning (AI/ML) to help scale decisioning—in other words, using decisioning to prioritize offers, deliver messages at the optimal time of day, select the best content for each person, and much more.
All of those steps lead us to design.
Once a brand has progressed toward, or even completed, the collection and curation of data, the organization must create a 360-degree view of all the messages required to deliver relevant content to all customers—aligned with their individual needs and across multiple channels. No big deal, right?
Actually, our agency took on solving this tremendous challenge a dozen years ago, and we’ve been diligently honing it ever since. Fortunately, it’s proven worthwhile in every way for our clients. Why? Because the 360-degree message view centralizes all of a brand’s content and connects it to delivery, decisioning, and measurement.
A key part of making that connection is thinking about the message as the fundamental idea being communicated and then linking it to all of its atomic content elements (e.g., headline, image, call-to-action, URL, etc.), presentation rules, and metadata needed to deliver it in any channel.
One message, multiple formats
Messages must be formatted differently for various placements—e.g., emails, web banners, direct mails, etc.—and manual formatting doesn’t scale. Any marketer who’s tried to do so understands the futility of this effort. Fortunately, breaking the message into atomic elements allows the structured content to automatically adapt to the visual and technical requirements of each placement—all while maintaining a consistent ID.
Moreover, because each placement is just a different presentation of the same atomic elements, any later edits you make are automatically applied to all instances and published to their respective channels—in a snap. This process makes it possible to create, maintain, and add messages in multiple formats and channels without any duplication.
Connect to decisioning and delivery
To get started, the organization’s decisioning engines need to know which messages are available. This can be done efficiently by capturing a segment, targeting metadata at a message-level, and then sending it in a standardized, machine-readable format to the decisioning engines. Here, message IDs are matched with individual customer IDs and fed to all channel-engagement platforms to deliver the right message to the right person, at the right time and in the right channel.
Connect to measurement
Next, to measure message performance across channels, the organization’s analytics platform needs to know how to identify the same messages. Manually creating and maintaining a map of messages in different engagement platforms isn’t practical at scale. However, naming messages with consistent IDs allows engagement data to be automatically connected and analyzed across channels.
How did we do it?
We started tackling this problem for our clients by customizing a commercial content management system (CMS). Eventually, we encountered obstacles and limits because we weren’t using the CMS for its intended purpose—managing websites. Our heavy customization made it challenging to get support or upgrade. When we couldn’t find a replacement, we developed an in-house solution from the ground up. It’s called CODA, short for Create Once, Deliver Anywhere.
And, wow, is it amazing to see all the atomic content elements automatically populate in messages with just the click of a button: The headline shows up where it belongs in the web banner, the Twitter promotion, the email, etc. In the same way, the image, body copy, call-to-action, and URL all slot in correctly within each unique message format. We use this high-tech solution for myriad clients and in multiple languages worldwide.
In brief, CODA is a power-up for organizations delivering personalized messaging at scale. Think of it as an add-on for advanced marketers, tying together their existing marketing technology, and making it possible to create and deliver consistent, customer-centric messaging at scale. In other words, it lets marketers deliver amazingly personalized customer communications. And because CODA is abstracted from individual channel engagement platforms, it enables consistent message creation, management, and measurement. What more could a brand ask for?
Adam Moore is vice president of Front-End Development and product owner of CODA at The Lacek Group.
Data Measurement: It’s Time to Level Up
Measurement is paramount in marketing today. Every decision we make as marketers to enhance efficiency and effectiveness must be measured before it can be assessed, either directly or indirectly. Without measurements, we wouldn’t know whether our decisions yielded the actions and results we hypothesized. In short, without measurements, we’re marketing with blinders on.
Measurements always start with business questions—e.g., “How is the overall channel performance?” and “How does the same message perform in various channels?” We want to see how every message, channel, campaign, and audience performs. Using channels as an example, we want to know the return on marketing investment, or ROMI, of the email channel, the direct mail channel, and the paid media channels. Plus, we want to understand the channel mix and attribution value of each channel to the final conversion rate.
More importantly, today’s precision marketing requires uncovering deeper insights into how interactions among channels, campaigns, audiences, and messages affect each other and overall marketing efforts.
Continuing our channels example, measuring channel-plus-message means considering the performance of the same message in various channels. These insights are critical—for example, we’d suspect that the same credit card acquisition message would yield largely different engagement performances on an e.com site compared to direct mail.
Similarly, measuring channel-plus-audience means studying the engagement of the same customer audience within various channels. As The Lacek Group learned from one of our client studies, the same audience can react differently to the same message, depending on if they come across it on a paid social channel or in their email inbox.
Ensure you’re realizing vital findings by:
1. Balancing measurements. Disproportionately large measurement efforts are put into assessing an individual channel or campaign performance and the attribution value among channels. That said, it’s the message that will resonate; customers don’t consider that they’re receiving a particular message because they qualify for a certain campaign or they’re on a certain channel. Precision measurement at the message level offers valuable opportunities for learning and optimization.
2. Using overlap intelligence. Brands often measure the performance of an entire campaign or communication. And that’s useful, but they can gain additional insights by measuring how a message performs differently within various channels or with several audience groups. In addition, while brands should measure the overall performance of a message, they can uncover more insights by analyzing how much that performance correlates with its delivery channel or audience group.
3. Focusing on customers. Brands often build in many micro-segmentations to precisely target particular customer groups. Interestingly, however, organizations rarely focus in on the performance breakdown at the micro-segmentation level. Create reporting that will surface useful segment information. Examples might include discoveries such as:
millennial customers’ engagement differs among channels
younger customers engage more than older customers with messages A, B, and C
female customers are more likely to respond to marketing messages in the morning while male customers are more likely to respond in the evening
elite-tier members engage less with social ads than infrequent customers
In short, despite our current era of customer-centric marketing, many measurement practices lag in their migration—they still prioritize the channel and/or the campaign rather than the customer.
What’s needed is a measurement framework that makes it easy to evaluate performance with granular precision and scalable pliability.
Using a measurement framework, brands must include balanced measurements and scale on overlap intelligence. Balanced measurement means giving equal measurement efforts to channel, campaign, message, and customer groups. Overlap intelligence means uncovering deeper insights into how interactions among channels, campaigns, audiences, and messages affect each other and overall marketing efforts. Balanced measurement and overlap intelligence are the start. From there, brands must improve their decisioning effectiveness.
How can brands do that?
Connect data. Build data pipelines—especially ones that complete the feedback loop. A brand must collect data on how customers respond and react to its served messages, and then feed that data back into the decisioning. That way, a brand can follow the performance holistically and end to end. After all, only connected data, not isolated individual metrics, ensures that a brand comprehensively follows prospects and customers through their journeys.
Build a content-tagging taxonomy, and implement it across channels. Tagging is the connective tissue that enables measurement. Achieving scalable granularity hinges on a smart tagging architecture, which should include essential information, such as customer audience, message or offer themes, communication or campaign, and channel.
Create and agree on measurement metrics across channels and teams. Measurement metrics can be categorized into a 3-E funnel: at the top of the funnel is Exposure, then Engagement, and at the bottom, Effect. Here’s how it works:
Exposure measures reach and awareness (e.g., match rate, number of impressions, number of deliverables, number of visitors, etc.).
Engagement measures interaction and participation (e.g., open rates, click-through rate, registration rate, number of sign-ups, number of downloads, etc.).
Effect measures business outcome (e.g., conversion rate, acquisition rate, cost per acquisition, incremental revenue, etc.).
Brands can use the 3-E metrics as a common language when discussing measurement among channels and teams. Additionally, measurement briefs can be created around each E, and each channel owner or team can log the individual metrics collected and measured.
Establish measurement reporting expectations. Measurement can be messy—often its roles and responsibilities aren’t clearly defined and communicated among teams. It’s important to establish and articulate measurement expectations around:
Ownership: Who produces and maintains the report?
Cadence: How often is the report refreshed (e.g., real time, daily, weekly, etc.)?
Visibility: Who has access to view and/or edit the report?
Delivery: How will the report be distributed (e.g., online dashboard, Excel attachment, PowerPoint slides, etc.)?
Recognize and leverage technology in the right order. Measurement should always start with business questions, not data or technology. Once the business questions are posed, apply technology to elevate measurement efficiency. Another savvy way to think about the role that technology plays is to adopt a new data technology to enforce and enhance measurement alignment among channels and teams. It’s often the most effective way to bring people and process together.
Measurement is the only way to know how our marketing investment performed. It’s the cornerstone of marketing optimization. And with the immense amount of data that we collect today, the expansion of marketing channels, and the integration among them (channels, functional teams, etc.), measurement tasks have become increasingly challenging.
Today’s marketplace is buzzing with new ways to synchronize and integrate performance data from various channels and sources into a single platform. In addition, as we shift our orientation of measurement frameworks from channel and campaign to customer and message, we gain new, scalable views into performance and optimization. And those views are worth all the hard work of today’s measurement strategies.
Shi Bu serves as senior vice president, of Data Intelligence at the Lacek Group.
Bringing It All Together
In this series we’ve taken a deep dive into The Lacek Group’s Personalization Engine. This technology-agnostic framework guides brands that want to plan and develop personalization capabilities and, today, nearly every brand does. Why? Because customers expect brands to understand their unique needs and expectations. By delivering messaging that recognizes and addresses personalization, a brand demonstrates its commitment to providing the experiences customers demand.
However, delivering message personalization at scale can be overwhelming given the coordination needed among data, technology, and business objectives. Employing a framework to meet that challenge is critical. Fortunately, our Personalization Engine breaks down this complexity into six stages.
This framework can help your brand deploy personalization inside your own organization. Use it to evaluate your current processes, capabilities, and technologies. From there, create a plan to pinpoint focus, prioritize technology investment, and coordinate team efforts. And, if you need help, please reach out. The Lacek Group is ready whenever you are to help rev up your personalization engine.
Julie Bustos serves as senior vice president of Marketing Innovation for The Lacek Group.
The Lacek Group is a Minneapolis-based data-driven loyalty, experience and customer engagement agency that has been delivering personalization at scale for its world-class clients for more than 30 years. The Lacek Group is an Ogilvy Company.