Shawn Reed, Blue Acorn iCi – Comprehensive & Accurate Data: The Cornerstone of a Successful Personalization Program

January 16, 2020

Learn how to deliver a more personalized experience to your customers with Shawn Reed from Blue Acorn iCi. He will discuss how to:

  • Build a solid foundation of data accuracy
  • Use comprehensive and accurate data to create powerful, personalized experiences

Fill out the short form to view the webinar on-demand.

Hi everyone and welcome to the Virtual Analytics Summit. I'll be talking to you today about comprehensive and accurate data, the cornerstone of a successful personalization program. Delivering a more personalized experience to your customers while also improving reputation and ROI requires comprehensive and accurate data. After all, a house is only as good as its foundation. So hopefully after today's session you'll have some new ideas to help you build that solid foundation of data and start taking your personalization program to the next level.

First things first, I'm Sean and I'm a hopeless data nerd. I even sometimes tweet about analytics and data stuff if you're into Twitter. Currently I am the senior director of marketing technology solutions at Blue Acorn iCi, we're a customer experience agency focused on delivering end to end customer experience expertise to our clients, including implementation of marketing technology platforms, content management systems, commerce systems analysis, data science, even product branding, packaging and shipping. That's right. We have our own warehouse, not what you normally would hear from a digital agency, but we're not your average agency. Over my career. I've done a number of different jobs including implementation consultant, developer, product manager, and yes, I've even done some QA, so I've had a lot of roles that I think are very relevant to the conversation that we'll be having today. 

So imagine a scenario for me for a moment. Let's say you go into your neighborhood grocery store and you walk up to the deli counter and you ask for a recommendation on a type of cheese that's good for melting. The deli employee says, "Sure I'll be right back." And when he comes back, for some reason he's holding a ham. You're very confused because you didn't ask for a ham at all. In your confusion and frustration at not being heard and understood, you turn and you walk away. The store employee calls after you, "But ma'am, you got a ham last time you were here so I thought you'd want another one. I'll be sure to note that you want cheese for the next time you come in!" But it's too late. You're gone and you can't hear him. A missed sale and perhaps a lost customer relationship all over a badly personalized experience. 

This might be a silly example and probably wouldn't actually happen in real life. However, scenarios like this happen all the time in the digital world when personalization is based on bad data. Don't let your company become the next to be famous for all the wrong reasons by delivering an embarrassingly bad personalized experience. Today I'm going to talk to you about why good data is so important if you want your personalization efforts to be successful. 

First, let's review the primary components of what I'll be speaking to you about today. These are the four primary things that I think are important to keep in mind when embarking on the journey of improving your company's data quality, and thereby improving your personalization capabilities and the experiences you're able to deliver to your customers and prospects. Understanding the impact of personalizing with bad data, putting data quality first, establishing an organization-wide digital center of excellence and setting realistic expectations about data quality and personalization.

Data quality can be kind of a subjective term and the application of it can vary significantly across different organizations. So I wanted to level set a bit and talk about what I mean when I say data quality. First, existence and comprehensiveness. Does your organization have all of the data that it needs? Does the data on a subject represent all of the data for that subject or is it incomplete? Second, validity and conformity. Are the values in your data set acceptable? Are they in the right format? Third, consistency. Does data across all your systems reflect the same information? Are there conflicts between the data anywhere that may create a problem? Fourth, integrity. Are the relationships being correctly maintained between datasets? Are all records connected as they should be or are there orphans? Fifth, accuracy. Does the data correctly reflect real world values? Are there any typos or misspellings in the data? And finally, relevance and timeliness. Is the data available when it's expected or needed? Is it kept up to date so that it's relevant to your organization's use cases? And here's a little hint to everyone. This is one of the most important components for personalization. After all, if your messaging is not relevant, it's not going to resonate and it might actually scare people away.

So let's start by talking about the impact of bad data. Forrester did a study that showed that 40% of consumers say most promotions they receive don't deliver anything of interest. This is a big deal because it means that a lot of time and effort and money is wasted on producing content that holds no value for the person receiving it. Oftentimes, this is because the content has been built based on incomplete information or poor assumptions, or maybe it wasn't personalized at all. In any case, this is a bad experience for the consumer and it's not likely to lead to success for your brand. 

in 2014 a well known online photo printing company mistakenly sent out emails saying, "Congratulations on your baby!" Unfortunately, most of the people who received that weren't actually pregnant. People who received the message understandably were confused, hurt, and even angry. They were left to wonder why this company would have chosen to send them a message that was not only irrelevant, but then in some cases evoked very negative feelings. Again, not a positive experience and one that likely drove a lot of consumers away from this brand. 

We've all had an experience where we visited a website or opened up a mobile app and had this look on our face. Why am I seeing this message? Are they intending this for someone else? Is this company worthy of my business if they can't even bother to get to know me? So let's look at some examples of bad personalization.

First, poorly targeted ads. Inaccurate or incomplete segments can lead to consumers being targeted for advertising based on traits that they don't actually possess. Like poor ad targeting, bad look alike modeling can result in a consumer receiving offers and experiences that make no sense because they're grouped with consumers who are not actually like them. Next, retargeting suppression or in this case lack thereof. We've all seen that situation where we get ads for something that we've already purchased. Sometimes it's even the same thing we already purchased through that very ad that keeps following us everywhere on the internet. And finally, wrong contact frequency. 45% of consumers say they receive too many offers and promotions. Knowing how often to contact a consumer based on their relationship with your brand is key. For example, if you know that a particular consumer is visiting your website daily and maybe even making purchases on a daily or weekly basis, you probably don't need to send them nearly as many emails as someone who hasn't visited your website in weeks or months.

Bad data is bad business and it creates added costs for organizations in a number of different and very frustrating ways. First, decrease conversion due to poor customer experience. This one is obvious, but it's still worth repeating. These days companies compete based on customer experience. In fact, 58% of consumers say that they've broken ties with a brand over a badly personalized experience. Next, poorly optimized marketing spend. If you can't properly attribute conversions, then you won't be able to optimize the allocation of your marketing investment across various channels. And finally, here's one that most people don't think about, decreased employee efficiency. Trying to gain insights from data that's incomplete or inaccurate creates a lot of frustration for analysts and mistrust from stakeholders And can be a huge waste of time and money. And that mistrust from stakeholders is a big one because most of us have probably experienced a situation where an executive in a company that we're working with doesn't trust the data anymore and therefore they're hesitant to support data driven decisions or support more investment in personalization. So this is a big deal here.

Another thing that should go without saying, but I'm going to say it anyway, don't be creepy. We've all had an experience for the brand where it seemed like they knew us a little too well. So if it seems like you know a customer a little too much, or if you're personalizing based on things that they viewed or did a long time ago, it starts to seem a little creepy. So what I recommend is to test and learn to strike a balance. Personalization is all about experimentation. Try new things with the experiences you deliver. But keep the consumer's point of view in mind. Would this experience delight you or would it make you feel a little uncomfortable? And keep legal requirements in mind. While it may be tempting to try and store every possible bit of information you can about all of your consumers, keep in mind that privacy regulations like GDPR and CCPA mean that you'll need to also be able to deliver a good experience even if you can't personalize because if someone opts out of data collection, you have nothing to personalize with. Therefore, store only what you know you need and use it with caution. 

All right, so let's go back to our definitions of data quality for a moment and look at them one by one. How can we fix bad data if we know we have it in our systems. First, with existence and comprehensiveness. If any records contain empty fields, you should do everything you can to obtain valid values for this field. Sometimes you can do this by checking other systems or you may even be able to incentivize the consumer to provide it to you voluntarily. Next validity and conformity. Run your database through a set of rules to check and ensure that values match defined criteria. For example, ensuring all phone numbers follow a common format. Next, consistency. Check various fields against one another in the same record or set up combined related records. For example, an age of 10 and a marital status of divorced on the same record probably indicates an issue. Next, integrity designing integrity checks for your data sets can ensure that new data brought into your consumer profiles is properly associated and not orphaned or siloed.

Next, accuracy. The best way to ensure accuracy is at the point of entry, but asking consumers to confirm certain key data points can be helpful in maintaining accuracy over time. And finally, with relevance and timeliness, examine the data you have in your database and determine if it's still applicable or if it's old enough to be purged or if you don't want to purge it, you can archive it in a way that it won't be used for personalization. So next, let's talk about putting quality first. Kissmetrics did a study that showed that the average enterprise organization can generate up to a 70% boost in revenue over other organizations based on data quality alone. That's a huge increase just from something seemingly as simple as ensuring you're using the correct data to build experiences. Data quality is definitely not achieved with a magic wand, as we'll talk about later, but hopefully the statistic helps you understand why making data quality priority in your organization is worth it. If you recall, back to my definition of data quality accuracy was one of the core components. Ensuring the accuracy of your data as it's collected is a great place to start on your data quality journey. Let's look at some different ways that we can ensure accuracy in the collection of digital data such as web analytics.

First manual unit testing. In my mind, quality starts with the developers that are responsible for building your brands experiences and implementing the tracking code that captures interactions with those experiences. So you should empower your developers with the knowledge to test their work and ensure that data quality is maintained as they make changes. If the developers don't understand how the analytics code works or why changing a certain variable breaks data collection, this is an opportunity for you to educate them on that. Next pre-production testing. Once changes have been made that have the potential to impact data quality, a full regression test of data collection should be performed in a preproduction environment to ensure that erroneous data collection doesn't negatively affect your critical production data.

And finally, automated testing and production. Depending on your content management platform, your release processes and other factors specific to your company, it may be possible that data collection can be broken in production even after it's been tested prior to a release. Because of this, it's very important to have an automated process to regularly test your data collection and production to ensure that any issues are identified quickly and can be resolved before bad data has a significant impact on the overall accuracy of your analytics and your personalization efforts. Tools like ObservePoint can be great at helping you automate data quality monitoring for your digital data. Other data sources in your organization such as financial data, will of course require other forms of monitoring and automation. 

So we talked a little bit about the accuracy, but what about the comprehensiveness? The more relevant data you're able to bring together, the better the experiences will be that you'll be able to deliver to consumers. Bringing together first party data from across your organization will help you to deliver more relevant and delightful consumer experiences. Deploying a customer data platform or CDP is a common way of combining multiple consumer data sets into a single consumer profile. Many options exist for this in the marketplace today, including Tealium and Adobe experience platform.

Next, we'll talk about the digital center of excellence. IBM did a study that showed that bad data costs the average enterprise between 10 and $15 million annually, which totals a staggering 3 trillion across the enterprises in the U$S alone. That is a huge number. How much is bad data costing your company? Most of us probably don't know exactly, but I can tell you that if you work to create a digital center of excellence in your organization, you'll be able to get a much better picture of that number and the things you need to do to significantly lower it.

So first, control access. Ensure that the ability to view and more importantly change your data is restricted to only those individuals who need it, and those individuals who understand the impact of any changes they might make. Next, be very careful with third party data. While it is valuable in personalization, you should use caution. A study showed that the average third party data can be up to 66% inaccurate. A preference should always be given when possible to first party data. Next, maintain consistency, creating standards for your data such as a common taxonomy used for consumer profile data across your organization goes a long way towards improving data quality. And establish roles and responsibilities. This is probably the most important one here. Determine a group of individuals in your organization who will remain committed to data quality and who will be accountable for ensuring quality is maintained over time. And before we move on, here's one bonus for you: get buy in from the top. It's much easier to get the support needed from a financial perspective as well as to keep data quality efforts from being deprioritized if the leadership of your organization also believes in the importance of data quality and knows that it must remain a priority,

The digital center of excellence should also be cross functional. It's absolutely critical to include members of all key business functions if you want it to be successful. And accountability and ownership should definitely be encouraged here. Groups like finance, legal, marketing, product, IT. and potentially others should all have a seat at the table and be part of this discussion. 

And don't forget to map out your journey. As you begin planning your digital center of excellence and discussing these plans with your colleagues across the organization, you'll naturally begin discovering sources of consumer data that you may not have known about. Use these discoveries to plot a course on how you will bring that data together in a way that is scalable, actionable, and then ensures the quality of the data. And finally, let's talk about setting realistic expectations. First, a quick stat from Forrester. 91% of marketers surveyed say they're prioritizing personalization, but many are still relying on basic segmentation or even rules-based targeting. This represents a huge opportunity for your company if you're willing to make data quality a priority. And if you're willing to think bigger in terms of how you can leverage your data to deliver personalized experiences, remember better data, means increased marketing opportunities 

So before getting started with your data quality journey, it's important to make sure that you're setting realistic expectations about your efforts and what they will look like so that you and others in your organization know what they're getting into. Companies are always trying to juggle a number of competing priorities and if it seems like data quality efforts aren't delivering on their promise, it's more likely that those efforts will be left by the wayside in favor of other things that are perceived to be more successful. So first, know that quality is never a one and done deal. As your business grows and evolves, your approach to quality must evolve with it and your data taxonomies must be kept up to date. Next, establish a baseline in order to demonstrate the return on investment of an ongoing quality program. It's important to understand where you're starting from.

So you can use tools like ObservePoint to do an audit, figure out what your data looks like today, what's working and what's not. Also, take a look at your KPIs, whatever's important to your business, whether it's conversion rate, average order value, etc., to know where those KPIs are before you begin your data quality journey. Then once you start making changes, you'll be able to see what impact those changes are making. Know that it won't happen overnight. Your data didn't become dirty overnight, so cleaning it up won't happen overnight either. Change takes time. And be willing to put in the work. Once you've made the decision to embark on your data quality journey, stay committed. It's tempting to deprioritize data quality for other things, but give into that temptation.

The key to making use of quality data and your digital experiences is to be willing to experiment and to start small. Using segments and personas can be one way to achieve this. Don't try to achieve one-to-one personalization from day one. Instead, focus on ensuring your consumers are grouped in meaningful ways and focus on the most valuable groups. Be bold and be flexible. Don't be afraid to try new things. While there are certainly best practices in the personalization world, it's also important to differentiate your brand. Channel, your inner empath. Put yourself in the shoes of your consumer. Is this an experience that would make you personally want to come back again and again? Data aside, personalization is ultimately about human connections. And finally, this is one of the most important rules of personalization, don't ask for what you have, meaning if a consumer has a reasonable expectation that you should already have the data, don't ask them for it again.

So I believe if you keep all of this advice in mind, ultimately your efforts will pay off. However, remember, it does take time. It will be worth it though. Companies fully invested in modern personalization will outsell outsell their competitors by 20% according to Forrester. So to sum up, by making data quality a priority in your organization, you can increase ROI, decrease marketing, spend, drive higher conversions, delight your consumers, and perhaps most importantly to some of the people listening to this presentation, you can make your analysts happier. So thank you for your time today. If you'd like to learn more about my company, here's a little bit of information and please enjoy the rest of the summit.

Previous Video
Erica Eischeid, Adobe – Creating a Data-Driven Culture Through Data Democratization
Erica Eischeid, Adobe – Creating a Data-Driven Culture Through Data Democratization

Learn from Erica Eischeid, Analytics Consultant at Adobe, how to foster a data-driven culture within your o...

Next Video
Adam Greco, Search Discovery – Reasons Analytics Implementations Fail (and How to Avoid Them)
Adam Greco, Search Discovery – Reasons Analytics Implementations Fail (and How to Avoid Them)

Adam will share what he has seen companies struggle with in web analytics and strategies to avoid these iss...