Here’s the summary (I’ve lined them up to each paragraph)
Businesses…
- still don’t have a clear customer view with everything about the customer
- haven’t needed to address the customer for some time — but times are changing
- have relied heavily on legacy approaches — but it can’t keep up with change and more information from more channels than ever before
- try to use all the data available to win over customers, but struggle to leverage all data — due to multiple systems, poor data quality, inconsistent customer view, inability to process in real-time, inability to engage when it matters
- also struggle because there are broadly two teams:
- Data teams that have specialist technical data knowledge
- Non-data teams that have specialist business domain knowledge
- … and there continues to be a disconnect between data and non-data teams — it’s slow and costly to turn data into actionable insight
- need to move quickly, but too often fall into the trap of picking a silver bullet solution, or attempt to restructure teams for better alignment with often limited success
- aiming to be customer-centric, need customer-centric infrastructure
- need to include new components of the customer data stack in their architecture, to empower all teams to operate independently AND collaboratively without hindering each other
A key enabler of this capability is the Customer Data Platform
This capability solves a number of people, data, technology and process hurdles, it introduces a new problem — premature communication — being too fast and efficient
There’s another article on exactly how we configured a real-time workflow end-to-end using Segment; the leading Customer Data Platform
Businesses still don’t have a clear customer view, with everything about the customer
Somehow, we’ve come to a point in time where after only a few short decades of trying to implement capabilities to treat customers as actual people — not just as accounts and transactions — we’re still all scratching our heads trying to understand who is the customer, what are they up to, how do they behave, what do they need/want and how do we engage with them right now?
Businesses haven’t needed to address the customer for some time — but times are changing
We’ve all, to some extent, surrendered to the almost globally accepted delusion that knowing who the customer is, is optional, and many organisations somehow continue to get by through sheer brute-force. This legacy has been guarded and protected for sometime — having been developed over many decades with significant investment — but like most things, it is reaching capacity and will expire. It has been failing to meet the needs of the organisation to truly understand the customer. Even data hub-based solutions (enterprise data lakes, warehouses and marts etc.) are falling short because they only provide some privileged and highly skilled teams with access to some data of varying degrees of quality to extract actionable insight — and this just takes far too long to deliver value.
Businesses have relied heavily on legacy approaches but can’t keep up with change and more information from more channels than ever before
With COVID-19 affecting many industries and businesses, particularly those that operate bricks-and-mortar stores, organisations are turning to and ramping up their online e-commerce capability to sell and engage customers — and with this influx of new digital channels comes a wave of richer customer data than ever before. It is imperative for organisations to have solutions to tap into this data as quickly and efficiently as possible, to extract deep, meaningful and actionable insight about customers to deliver better customer experiences.
Businesses try to use all the data available to win over customers, but struggle to leverage all data — due to multiple systems, poor data quality, inconsistent customer view, inability to process in real-time, inability to engage when it matters
From our many years working with customers, the story about engagement typically goes like this: For each customer, the business is tapping into historical transaction and customer record data, with some highly aggregated data points from external sources (like market research, social, census demographics etc) and process this as best as possible, leveraging perhaps some machine learning capability to classify customers and predict their value, probability to churn or likelihood to be interested in product promotions. Many also try to leverage data from real-time channels like digital tracking (think Google Analytics) to build audiences based on current customer interactions and journeys, and this is often done within those systems and is rarely blended with all the other available enterprise data. And even if both real-time digital and batch enterprise data were available in say a warehouse or lake, it’s likely to be too slow to come together to take action in moments that matter.
In addition to the challenges of bringing the data together in one place, it’s quite common to hear that the data is rarely of suitable quality for analysis, or that it’s not appropriately structured and anchored to the concept of a customer (the real person) — and still linked to non-customer subjects like accounts, households, services, billing entities etc. Furthermore, the data is so rich and detailed, that to find meaningful signals and patterns of behaviour — you’d need to be a specialised data analyst/engineer/scientist with solid business domain understanding to extract that useful information for product and marketing teams to use — and again, this is usually a bottleneck. The whole business wants to access that rich actionable insight and information, and disseminating it to where it’s needed is too hard, or often in the form of a visualisation or dashboard that only answers some questions and is limited in its ability to action the insight.
Businesses also struggle because there are broadly two teams: Data teams with specialist technical data knowledge and non-data teams with specialist business domain knowledge
If I try to oversimplify the problem, most organisations typically have two broad types of teams in various organisational structures:
- the Data Whisperers — those that have the technical expertise to wrangle data to extract actionable insight
- the Data Dreamers — those that run the business (sales, marketing, product, risk, customer service, customer experience) that need actionable insight
There continues to be a disconnect between data and non-data teams — and it’s slow and costly to turn data into actionable insight
Data Whisperers often lose sight of what they’re here to do — serve the business. They don’t exist without them and their purpose is to solve business problems and add value, not just play with new tech.
The Data Whisperers are technically trained and experienced to structure and process data to derive information and actionable insights, leveraging approaches and technologies like SQL, statistics, machine learning and artificial intelligence — which are completely outside the realm and capabilities of the rest of the business. Data Whisperers rely on requirements from the business to turn over insights, but more often than not, the requirements are vague, or cannot be actioned in time or at all.
Meanwhile, the rest of the business, the Data Dreamers who, in addition to fighting and putting out daily fires responding to volatile external and internal forces, are also trying to stay aligned to strategic financial and customer experience targets and desperately need data, information and actionable insights to keep on top of things. The Data Dreamers need to be able to self-sufficiently analyse enough data to a less sophisticated but equally important degree than the Data Whisperers, so they can actually action the insights with speed and accuracy to keep up with the volatility of the business.
Traditional data marts, visualisation and dashboard just won’t cut it anymore as the only solution here — because insights embedded in visualisations and dashboards stop flowing. It turns into a manual intervention and process to interpret the insight, return to the Data Whisperers to get access to the underlying data and then manually take action to do something about it. And that assumes that you have a dynamic enough visualisation and dashboard to answer any and all questions as they come to you — which is rarely the case. Recently, solutions leveraging Data Notebooks have taken hold, as this is far more rapid, agile and dynamic than visualisation tools alone — but the challenge here is that it still relies on a technical person to wrangle the data.
You’ve almost certainly experienced this: the business needs an answer to a relatively simple question about customer experience that isn’t available in any current visualisation or dashboard, and reach out to the data and analytics team with requirements to get the answer to the question. They come back a few times, having found it difficult to understand the requirements or source the data for the problem, and respond with a heart-breaking “that’ll be a million bucks and take 3 months”. And the business team all look at each other with defeat, knowing that it’s completely infeasible, and by the time they get it the opportunity will have passed and it will likely be wrong anyway, or by then they’ll have a new question — one they hadn’t even considered yet.
Businesses need the ability to move quickly, providing appropriate control over customer data for different teams
Today, businesses need to move quickly and provide real-time control over customer data; to support the business in understanding who they are, how they behave and what they’re likely to do next. And this needs to be available to both data and non-data teams simultaneously.
Some look to single vendor shiny, silver-bullet monolithic tech stacks that are difficult and costly to deploy, aren’t as agile, dynamic or function-rich and customisable as they’d hoped and make it difficult to calculate or realise a return. Some continue to centralise the entire data and analytics function, and end up with a task log bottleneck that never gets fulfilled. Some try to decentralise the data and analytics function completely, and end up with distributed systems and people of varying skills and capability not following standards embedded in teams with custom definitions of key business concepts like Customers and Accounts.
Customer-centric organisations need customer-centric infrastructure.
The modern customer data stack provides a range of out-of-the-box and customisable services that are fully integrated with open APIs and play really well together, providing a specific suite of capabilities that permit both data and non-data teams to leverage data and analytics to identify and activate key engagement opportunities at every stage of the customer journey.
The modern customer data stack can:
- Tap into and integrate many data sources — out-of-the-box
- Process data in real-time or near real-time
- Resolve customer identities and profiles across data sources, with simple configurations
- Dynamically enhance and enrich customer profiles, as new systems and data comes available
- Provide control to non-technical/non-data business teams to create insight, for analysis and activation — easily, quickly and self-sufficiently via simple, intuitive interfaces
- Provide control to data teams to easily augmenting customer profiles, with deep, rich and advanced insight from existing data hubs and advanced analytics (AI/ML) solutions
- Send raw data and actionable insights into data warehouses, data lakes, big data and analytics platforms and more, to provide real-time signals for advanced analytics
- Send actionable insights into many target systems — sales, marketing, social, advertising, personalisation, analytics systems and more — to activate engagement with customers in real-time
- Maintains a high degree of governance and security
With this solution at your fingertips — just imagine the possibilities. And it’s here, today.
In the following three diagrams, I provide a very high-level overview of the components of the modern customer data stack that can deliver all of the above capability.
In this first version of the diagram, you can see how each part of the customer data stack satisfies different needs of different business units.
In this second version of the diagram, you can see how each part of the customer data stack is typically labelled according to its target problem domain.
In this final version of the diagram, you can see each of the individual example components as they come together to provide the business full and multi-speed control over customer data to fulfil the various requirements of both data and non-data teams, independently and collaboratively, without hindering overall momentum or effort.
For those really itching to learn more about each of the components, I suggest this further reading:
- Customer Data Platforms (CDP) — Segment
- Analytics Platforms — Mixpanel
- Marketing Automation Platforms — HubSpot
- Customer Engagement Platforms — Braze
- Cloud Infrastructure, Big Data and AI/ML services — Google Cloud
So… While I hope you found the above content useful — you’re probably wondering “when’s he getting to the punch-line” and “what new problem is this creating?”.
When I personally went about implementing this capability (yes, I promise you it’s real and not just fluff and you can see it done here) — by integrating and configuring the various tools of the modern customer data stack as illustrated — I realised that because it was so easy, efficient and straightforward to build real-time workflows among all the platforms I needed for customer engagement, I was now completely at risk of communicating prematurely!
In just a few clicks, I can identify certain types of behaviours of interest within my customer base. I can create custom traits/signals to flag those customers in real-time when they behave a particular way. I can create an audience of customers who behave or don’t behave in that way — in real-time. And I can send that audience to my Marketing Automation system of choice to shoot them a campaign… yes… also in real-time. When you consider that most of this effort is just configurations in various GUIs (no programming or complex logic) and the real-time nature at which data flows through these systems, it can be quite daunting in some cases I suppose for the customer to be communicated to so quickly — that you might end up introducing intentional “wait-times” in your workflows to delay the delivery of communications and other actions, to avoid communicating so quickly or perhaps prematurely and appear less… well… stalky…
I told you you’d get a problem… But c’mon… what a problem to have!!!
In another article, I’ve published the steps I took to configure an end-to-end workflow across two technologies of the modern customer data stack, Segment and HubSpot, which in the past would have taken forever to do, with a tonne of customisation and bespoke integrations. From the moment data was flowing from our custom-built Online Portal into Segment, it took me about half a day to configure a very specific workflow that sends an email via a HubSpot Marketing Campaign to anyone that searched for me on the Online Portal. And it would send the email almost instantly after the event. If I can deliver something that specific, in just half a day… trust me… there’s no limit to what you can do with this stack and bit of creativity.
