We all know the bottom line: to successfully cultivate value-based customer relationships, we need to fully know our customers. And that means creating a foolproof SCV―Singular Customer View―that aggregates and unifies all the information we can glean from the vast amount of information that every business organization generates and collects.
Easier said than done, of course. With varying degrees of success, I have been struggling to build comprehensive and accurate SCVs for over fifteen years. Complete data-driven customer profiles lie at the core of insight-driven, one-to-one client communication, but I must acknowledge that creating them is a necessary evil. Achieving a stable and steady SCV is a constant grind: a bother to build, a challenge to maintain, and a drag to sustain.
Simply put, building an SCV requires identity resolution, whereby data points from multiple data sources are matched and merged to construct a persistent and accurate view that contains profile, behavioral, transactional, and demographic points of reference. It’s a complex minefield with all sorts of potential problems with data hygiene, data inconsistency, data sprawl, and the challenges of system integration. ID resolution also requires mastering a myriad of unique identifiers such as mobile IDs, device IDs, cookie IDs, and IP addresses, along with known identifiers such as email addresses, phone numbers, and so on.
But today, there is cause for optimism. The dawn of Artificial Intelligence and Machine Learning methodologies, AI/ML, is revolutionizing the way we connect and understand customer data. As a result, today we are seeing astonishing progress in identity resolution. AI/ML identity resolution solutions help to tame an organization’s data chaos and build meaningful customer datasets that enable markedly better, insights-driven customer experiences. These solutions allow for more personalized, value-based interactions with customers, making them a wise choice for any organization to consider today.
In the past, there were few out-of-the-box solutions available to build an SCV. Identity resolution was a manual, laborious process that was prone to errors and omissions. Generally, the process entails onboarding data into a tabular on-premises database and then writing lengthy and complicated SQL codes to match (using one-to-one matching) and merge unique identities across data sources. However, due to the restrictive nature of the tabular format, it was challenging to handle data volumes, variety, and velocity efficiently, and this often led to problems with scalability and data accuracy. This manual solution was clunky and limited in its capabilities, and it crashed all too often. Moreover, no matter how clean the data was, this approach was unable to account for web browsing data, device hopping data, paid media interaction data, customer service data, and other problems.
It was also hard to identify duplicate profiles and merge them together. For example, a common issue when using a rule-based approach in identity resolution is that if a customer has two email addresses, it appears that there are two different customers when in fact, there is only one. Despite all the time, money, and resources invested in building and using an SCV, there was a constant need to troubleshoot and resolve data glitches. The resulting lack of trust in the system meant that businesses weren’t confident in relying on it for their data needs, creating an obvious barrier to productivity.
Today, AI/ML identity resolution is dramatically changing the landscape―offering a powerful, automated, and flexible process to help organizations achieve optimal SCV. This process uses advanced algorithms to identify and match customer records across multiple data sources and data points, a methodology that offers greater flexibility, accuracy, and scalability, and that can handle complex and evolving match-and-merge scenarios.
AI/ML identity resolution offers multiple advantages over rule-based approaches, among them:
1. Greater Accuracy: AI/ML identity resolution algorithms are more accurate than the rule-based approach since they can continually learn from data patterns and adjust/improve match-and-merge criteria automatically. This results in fewer false positives and fewer false negatives, which enhances the accuracy of SCV.
2. Scalability: The AI/ML approach can handle large volumes of data with ease, which makes it more suitable for organizations that generate such data volumes.
3. Flexibility: This methodology can handle different data types with little data transformations and automatically adapt to changes in data from data sources.
4. Better handling of complex scenarios: Algorithms are proven to handle complex match scenarios such as partial matches, fuzzy matches, and multiple matches more efficiently than traditional approaches. Furthermore, these algorithms continually improve over time rather than being static.
5. Reduced maintenance: AI/ML identity resolution algorithms require less manual intervention and maintenance in the long run, which results in lower costs and reduced complexity.
I’ve repeatedly been asked whether the ID graph methodology can do the same thing that the AI/ML approach does. It can't, and it doesn’t. True, the ID graph methodology is an improvement over the rule-based approach because, instead of traditional tabular and relational databases, it uses graph databases to represent customer identities and their relationships. So yes, graph databases provide more flexibility in linking across multiple data sources. But the ID graph methodology still relies on humans to define match and merge rules based on specific unique identifiers (such as email addresses, phone numbers, etc.). This means that the ID graph methodology will always be limited by the accuracy and completeness of how fallible humans apply the rules.
As I once did, many of you might have already invested in ID resolution solutions. As a result of the rapidly evolving AI/ML landscape, do not be reluctant to re-evaluate those solutions to ensure that your current investments are delivering on the promise of value-based personalization. It may be time to consider exploring more modern AI/ML-based ID resolution solutions, since these are likely to be more accurate, efficient, and effective than the approaches you currently have in place.
Evaluating your ID resolution options these days is a tricky business, and you may be tempted to rely on vendors’ representations about how their sausage is made. Everyone is going to claim perfection. Don’t bite too fast or too hard. Instead, buckle down and do your homework. There is a plethora of vendors in the marketplace offering ID resolution solutions, and that number keeps growing. However, only a handful of them truly applies apply AI/ML methodologies. While most vendors can speak eloquently about what their ID resolution offerings deliver, only a handful can articulate how they solve for it. There’s a lot of snake oil out there. Be sure to do your due diligence when selecting a solution and take advantage of proof-of-concept opportunities to ensure that you understand “how" these solutions meet your needs.
Moreover, technology that is novel and marvelous this week may be obsolete, or at least ho-hum, next week. So, in addition to being smart about your current decision, it behooves you to keep abreast of news about how tomorrow’s sausage is going to be made. As Winston Churchill reminded us, “perfection is the enemy of progress,” something particularly true in the world of ID resolution.
At Blend360, our focus is on helping our clients develop and implement effective customer data strategies that build a stable, persistent Single Customer View and drive personalization. As a specialist in Customer Data Strategy, I am confident in our approach and our ability to customize it to meet the unique needs of each client. If you're interested in learning more about how we can help you optimize your customer data strategy, please don't hesitate to get in touch with us here.
We all know the bottom line: to successfully cultivate value-based customer relationships, we need to fully know our customers. And that means creating a foolproof SCV―Singular Customer View―that aggregates and unifies all the information we can glean from the vast amount of information that every business organization generates and collects.
Easier said than done, of course. With varying degrees of success, I have been struggling to build comprehensive and accurate SCVs for over fifteen years. Complete data-driven customer profiles lie at the core of insight-driven, one-to-one client communication, but I must acknowledge that creating them is a necessary evil. Achieving a stable and steady SCV is a constant grind: a bother to build, a challenge to maintain, and a drag to sustain.
Simply put, building an SCV requires identity resolution, whereby data points from multiple data sources are matched and merged to construct a persistent and accurate view that contains profile, behavioral, transactional, and demographic points of reference. It’s a complex minefield with all sorts of potential problems with data hygiene, data inconsistency, data sprawl, and the challenges of system integration. ID resolution also requires mastering a myriad of unique identifiers such as mobile IDs, device IDs, cookie IDs, and IP addresses, along with known identifiers such as email addresses, phone numbers, and so on.
But today, there is cause for optimism. The dawn of Artificial Intelligence and Machine Learning methodologies, AI/ML, is revolutionizing the way we connect and understand customer data. As a result, today we are seeing astonishing progress in identity resolution. AI/ML identity resolution solutions help to tame an organization’s data chaos and build meaningful customer datasets that enable markedly better, insights-driven customer experiences. These solutions allow for more personalized, value-based interactions with customers, making them a wise choice for any organization to consider today.
In the past, there were few out-of-the-box solutions available to build an SCV. Identity resolution was a manual, laborious process that was prone to errors and omissions. Generally, the process entails onboarding data into a tabular on-premises database and then writing lengthy and complicated SQL codes to match (using one-to-one matching) and merge unique identities across data sources. However, due to the restrictive nature of the tabular format, it was challenging to handle data volumes, variety, and velocity efficiently, and this often led to problems with scalability and data accuracy. This manual solution was clunky and limited in its capabilities, and it crashed all too often. Moreover, no matter how clean the data was, this approach was unable to account for web browsing data, device hopping data, paid media interaction data, customer service data, and other problems.
It was also hard to identify duplicate profiles and merge them together. For example, a common issue when using a rule-based approach in identity resolution is that if a customer has two email addresses, it appears that there are two different customers when in fact, there is only one. Despite all the time, money, and resources invested in building and using an SCV, there was a constant need to troubleshoot and resolve data glitches. The resulting lack of trust in the system meant that businesses weren’t confident in relying on it for their data needs, creating an obvious barrier to productivity.
Today, AI/ML identity resolution is dramatically changing the landscape―offering a powerful, automated, and flexible process to help organizations achieve optimal SCV. This process uses advanced algorithms to identify and match customer records across multiple data sources and data points, a methodology that offers greater flexibility, accuracy, and scalability, and that can handle complex and evolving match-and-merge scenarios.
AI/ML identity resolution offers multiple advantages over rule-based approaches, among them:
1. Greater Accuracy: AI/ML identity resolution algorithms are more accurate than the rule-based approach since they can continually learn from data patterns and adjust/improve match-and-merge criteria automatically. This results in fewer false positives and fewer false negatives, which enhances the accuracy of SCV.
2. Scalability: The AI/ML approach can handle large volumes of data with ease, which makes it more suitable for organizations that generate such data volumes.
3. Flexibility: This methodology can handle different data types with little data transformations and automatically adapt to changes in data from data sources.
4. Better handling of complex scenarios: Algorithms are proven to handle complex match scenarios such as partial matches, fuzzy matches, and multiple matches more efficiently than traditional approaches. Furthermore, these algorithms continually improve over time rather than being static.
5. Reduced maintenance: AI/ML identity resolution algorithms require less manual intervention and maintenance in the long run, which results in lower costs and reduced complexity.
I’ve repeatedly been asked whether the ID graph methodology can do the same thing that the AI/ML approach does. It can't, and it doesn’t. True, the ID graph methodology is an improvement over the rule-based approach because, instead of traditional tabular and relational databases, it uses graph databases to represent customer identities and their relationships. So yes, graph databases provide more flexibility in linking across multiple data sources. But the ID graph methodology still relies on humans to define match and merge rules based on specific unique identifiers (such as email addresses, phone numbers, etc.). This means that the ID graph methodology will always be limited by the accuracy and completeness of how fallible humans apply the rules.
As I once did, many of you might have already invested in ID resolution solutions. As a result of the rapidly evolving AI/ML landscape, do not be reluctant to re-evaluate those solutions to ensure that your current investments are delivering on the promise of value-based personalization. It may be time to consider exploring more modern AI/ML-based ID resolution solutions, since these are likely to be more accurate, efficient, and effective than the approaches you currently have in place.
Evaluating your ID resolution options these days is a tricky business, and you may be tempted to rely on vendors’ representations about how their sausage is made. Everyone is going to claim perfection. Don’t bite too fast or too hard. Instead, buckle down and do your homework. There is a plethora of vendors in the marketplace offering ID resolution solutions, and that number keeps growing. However, only a handful of them truly applies apply AI/ML methodologies. While most vendors can speak eloquently about what their ID resolution offerings deliver, only a handful can articulate how they solve for it. There’s a lot of snake oil out there. Be sure to do your due diligence when selecting a solution and take advantage of proof-of-concept opportunities to ensure that you understand “how" these solutions meet your needs.
Moreover, technology that is novel and marvelous this week may be obsolete, or at least ho-hum, next week. So, in addition to being smart about your current decision, it behooves you to keep abreast of news about how tomorrow’s sausage is going to be made. As Winston Churchill reminded us, “perfection is the enemy of progress,” something particularly true in the world of ID resolution.
At Blend360, our focus is on helping our clients develop and implement effective customer data strategies that build a stable, persistent Single Customer View and drive personalization. As a specialist in Customer Data Strategy, I am confident in our approach and our ability to customize it to meet the unique needs of each client. If you're interested in learning more about how we can help you optimize your customer data strategy, please don't hesitate to get in touch with us here.
Lorem ipsum dolor sit amet consectetur. Sit sit vel eu sed proin. Arcu bibendum in mauris urna erat magna volutpat eget. Lectus commodo consectetur egestas quis ut lobortis nunc.
.png)
Lorem ipsum dolor sit amet consectetur. Sit sit vel eu sed proin. Arcu bibendum in mauris urna erat magna volutpat eget. Lectus commodo consectetur egestas quis ut lobortis nunc.
.jpeg)
