As already mentioned in the first part of this blog series, we often get the reaction of organizations that they’re already working on CX or that they’re quite ‘good’ when it comes to it. They managed to leverage customer data for providing a relevant interaction. However, we notice that the company’s internal expectations are rarely met. A pilot was delivered, but they are struggling with the next steps. We often hear they expected to see results sooner, but that the first use case still isn’t on point months after the launch, or that they have little overview of the various use cases and how they impact each other. Next to that, the companies we talk to realize they need a lot more people for the setup and maintenance of the CX program department-wide, exceeding the initial budget and putting the ROI under pressure.
With all of this, the scalability of CX is a serious challenge to overcome. Digging deeper, we can distinguish 5 typical CX scalability issues: setting up an extra use case, maintaining the implemented use cases, orchestrating the various use cases, the infrastructure beneath the platform and the team members that have to work on the use cases. Ignoring these is not an option. You should see this as a checklist to figure out in which areas you’re already doing well and where you need to put in extra focus and effort.
In the first part, we already elaborated on the 3 first issues. In the following section, we’ll tackle the 2 remaining issues and we’ll finish with some basic principles that will help you to conquer those challenges.
There are basically two types of infrastructure issues when it comes to CX scalability. On the one hand, you have architecture scalability and on the other, you have big data scalability.
With architecture scalability, we refer to the storage, or processing, capacity you have. This capacity system – both on-premises software and in the cloud – has to be easily scalable. But with physical on-prem software, the amount of data you can render, calculate, use or add, depends on the physical capacity of the hardware.
This capacity is often sufficient in the early stages, but the future ambitions are usually a lot greater than what the architecture can take. Now, who already thinks about what future capacity he wants the moment he invests in the hardware? That’s right, yet another challenge. Luckily this is mostly an on-prem issue, as it goes a lot easier when you store data in the cloud. Next to this, we often see there are a lot of IT resources needed for it, something you don’t want of course. We say the best way to go is to choose a data platform where you can easily add data to, without having the storage restrictions.
The best way to go is to choose a data platform where you can easily add data to, without having the storage restrictions.
Big data scalability
You need to be able to both add new data sources easily and quickly and to add an unlimited amount of data (per source) at the same time. Working with all kinds of structured and unstructured data, it is important to choose a real big data platform that carries this native in its core.
You have your current existing data sources, but there are also future data sources. Key here is that the infrastructure should let you connect those easily, based on technical components that allow you to build, process and monitor future data flows in and out of the system.
New setups often depend heavily on the input of hard skills like IT experts or data scientists. That expertise is needed every time a new use case is developed in order to make changes, to code the new scripts, etc. This human factor absolutely limits scalability too. Needless to say that more people mean a higher cost and new team members require a longer onboarding time when they’re dealing with very complex use cases. That is if you can find those profiles at all, to begin with. In Europe, 47% of organizations are struggling to fill data science positions. One solution for this war for talent is to empower the people who are closest to the customer: marketers and client service representatives, and encourage them to become so-called citizen developers. Next to this, usually, only key personnel has an overview of all the use cases, something that can get very messy when those people leave the company.
In the beginning, of course, you will need experts to tie the data together, to clean it, to code it, and whatnot, but after a couple of use cases, empowering the marketer is key. This way, things will move a lot quicker as the marketer will need the IT support team way less after a while. With a good centralized data platform, the marketer can for instance now create his own audience groups, in a very detailed way. It could then be an audience with people (1) between 25 and 35 years old, (2) who used the credit card simulator in the past 2 weeks, but (3) still haven’t opened a credit account until this day, and (4) who are eligible for a certain promotion and experience, such as a personalized email informing them what they’d on average save in accordance with their age category and life context. It is important the CDP has a very intuitive UI to make the transition as smooth as possible, and make sure it has a standard inclusion of the most important machine learning applications.
New setups often depend heavily on the input of hard skills like IT experts or data scientists. But with a good centralized data platform, the marketer can take this into his own hands.
No longer is the ability to use technology to create offers from products and services or fine-tune and target their attributes and their promotion restricted to a highly trained technical elite. Now citizen developers can use building blocks to create new offerings.
How this impacts your CX
If you don’t take the issues from above seriously (read the first part here), your customers won’t have the ultimate experience you want them to have when coming into contact with your company.
Imagine you’re a media company and you have 1 use case that delivers experiences to people interested in a sports deal. But what about the people who are interested in movie classics? They should also get the right messages at the right time. So if it’s a burden to add that second use case, it obviously leaves a lot of opportunities and this will impact the CX. Without a central platform that runs your use cases automatically, your customers won’t receive relevant messaging as they’ll see alternating offers. As such, they won’t have a consistent experience throughout their customer journey. Also, when you don’t have a full overview of your audiences, you might overlook a subset of customers and some groups may get a better CX in the end.
When the technical maintenance goes wrong, and the data sources change the format of the data, for instance, your customers could end up being addressed by their last name instead of their first name or they might just receive completely wrong information. Even worse is when you’re spamming your customers without even knowing it when orchestrating between multiple use cases. And when something goes wrong with the maintenance of the use cases, a CDP will not only alert you, but it will at least always choose a basic nurturing campaign to fall back on.
Your CX might get stuck due to poor performance. You could only have 5 to 10 use cases, and only see the number of personnel growing instead of extra CX use cases. Best case, you have a single point of view on all the use cases, however, usually, no one will have this overview. This leads to situations of sending your customer 37 messages over a period of 14 days during the onboarding trajectory without even knowing as they all came from different departments.
Some key tips
Basically, the scalability of customer experience can be limited by many factors. It is important to take into account the five above elements (here is part I of this series), and at the beginning, they may seem like big hurdles to overcome. Here are some key points of advice, things you might want to consider before diving deeper into it.
1. Start small
Start small with a few data sources and a simple data model for quick ROI. You typically start with only 3 data sources. The most common are your CRM and the customer’s website and mobile behavior, as they are the easiest to implement and they already give you a good client profile. Choose the properties you need for your first use case, and stick to only those. Also, focus only on a limited number of execution platforms such as the website/CMS and email. A first use case could be a website banner you show to the right customers and a consecutive email campaign to follow-up on it. This way, you can already start immediately with the resources you own and you’ll see a quick ROI without needing a long implementation time. After this, you can add more use cases, execution platforms, data from the CRM, etc.
2. Infrastructure scalability
Make sure the data platform you choose supports the infrastructure scalability. Big data requires a distributed architecture. And it’s not just about storing data, but also about accessing, integrating, processing and performing calculations on data. And then there’s also the duality between on-prem software and cloud storage. The first option comes with physical restrictions, while the cloud comes with the distinction of shared versus private storage. With a private cloud, we set up 1 instance for 1 client so it can never get mixed up with other companies’ data.
Financial services and banks are of course very reluctant to share cloud space with other instances. But on prem-software is, as we’ve already mentioned, not scalable at all, so you need to estimate the storage capacity upfront. Consider all the options before making a decision on this.
3. Empower the marketer
In order to define audience groups from your customers in the past, the marketers needed the help of data analysts. This was always a very time-consuming process and it took a while to get the results you asked for. Give your marketers the right tools so they don’t need the support from the CRM team or IT team when building out offers and experiences to provide a CX. They need to be able to define their own targets and have the ability to access metrics to obtain new insights, discover opportunities and audiences, and act on them swiftly.
Moreover, marketers should also be able to use the tool to gain insights on how they are successful in providing meaningful interactions that deliver value to the customer and where not. This is an ongoing process of fine-tuning and continuously improving to make sure the customer expectations are met or even exceeded when it comes to the products and services provided.
Building integrations doesn’t always happen smoothly. Over time, you’ll need to add other channels, those are either the sources from which you assemble data or the channels on which you’ll reach your customers, such as the social media platform TikTok. There are a couple of technical ways to connect data, such as through APIs or an Amazon S3 bucket, so you need to be able to build a bridge between the existing data and collecting new data. This comes with some work, as you need to: schedule the data collection process, determine the format you want it to be collected in and monitor the uploading of that data.
You need to find a solution that helps you connect with new data sources very quickly and easily. With that, you also need to be there where you have a connection with your customers in order to contact them quickly over those channels and touchpoints.
5. Change management
Now, we know you can’t change the internal company structure overnight, but it is, however, essential when you want to achieve your ROI in the near future. We often see that the various channels have specific owners, each with their own Profit & Loss accountability. At some companies, we see that even the various sub-branches have their own P&L. A multi-touch attribution model is the best solution for this as it divides credit to the channels, campaigns, keywords, and touchpoints that contributed to the sale. Read more about that in this article.
Your marketing department, for instance, should have an omnichannel approach instead of being spread out across separate channels. This goes hand in hand with a one-on-one focus instead of segment-based marketing. Putting the customer first means delivering relevancy and empowering the customer to make decisions in life.
Many organizations struggle to pursue a shift in customer experience in combination with their day-to-day operations. Begin with appointing cross-functional teams that work within their existing roles and in parallel with reporting structures (McKinsey & Company). Adopt an agile approach and quickly identify customer pain points, create minimum viable products and have them tested rapidly. Iterative testing lets your team test new prototypes quicker, they learn from the failure and feedback and then refine until it’s a success. As McKinsey & Company stated: “Over time, organizational structures can be revamped to deliver the new experiences sustainably over multiple years. The result is a transformation that delivers early impact and momentum and an opportunity to evolve as needs change, without the disruptive shock of tearing up an operating model in the fragile, early stages.”
We know you can’t change the internal company structure overnight, but it is, however, essential when you want to achieve your ROI in the near future.
Where can NGDATA help?
NGDATA empowers you to better serve your customers. Our Intelligent Engagement Platform (IEP) is much more than a regular Customer Data Platform (CDP) as it goes beyond unifying all your customer data and allowing customer segmentation.
Supported by our AI capabilities, the business user can analyze, predict, and orchestrate dynamic 1-1 experiences across millions of customers in real-time, on the preferred channel in a relevant context. Because personalized, one to one marketing is the only way to build profound relationships that drive customer retention.
Dig deeper with us to see how you can overcome these scalability issues.