Data onboarding generally means bringing in data from some outside source into some specific system.
When we talk about customer data onboarding, we’re talking specifically about taking data from your customers and loading it into your system or platform where you can work with it and return something of value to those customers.
The most common scenario we see is a SaaS platform needing to onboard data to get a new customer up and running – and there’s usually also a need for some ongoing data ingestion too, so the customer is seeing their up to date data in the platform.
Why is customer data onboarding so important?
"The onboarding process is our first chance at making a good impression with our customers after the dollars are already spent"
- Bryan Kahlig, Senior Director Product Development, Zywave
How you handle data onboarding gives a critical first impression of you as a vendor. And this can happen even before you’ve signed your customer. As part of your pitch you want to be able to give reassuring and positive answers when prospects are asking questions about your onboarding process such as:
- How painful is the data onboarding process going to be?
- How long is it going to take?
- How much work will I need to do?
- What will I need to do every month, or week, or day, to make sure my data is up to date?
And once you’ve signed the deal, you want to make sure that good impression carries on.
Of course the onboarding process may be iterative, requiring adjustments or re-mapping, but the crucial thing is that this should be as transparent to the client as possible. A lack of transparency, unexpected delays or lots of errors can leave your new customers asking themselves ‘did we make the right decision?’
The speed and efficiency of the onboarding process is also valuable to you as the ‘onboard-er’. If your engineering team is having to do a lot of time-consuming, manual work to onboard each new client, that not only costs you money, but also means you may not be able to keep up with the number of customers you need to onboard. These bottlenecks not only hinder your scalability but can also leave your new customers frustrated with the time it takes to get them live.
Challenges of customer data onboarding
The #1 challenge with onboarding data from multiple clients is that you’ve got no control over the format or quality you receive the data in. In an ideal world, you want to be able to take whatever they have easily and get it into your system.
"When I go in and I speak to a prospective customer, I don't ever worry about data. I did before. The question was always ‘Where's your data? What does it look like? What format is it? How much are we talking?’
So what we were doing before was we were kind of going in with handcuffs. And what CloverDX really allowed us to do is go in and say ‘It doesn't matter how you're giving us this stuff, we're just going to stitch it all together.’"
Other data onboarding challenges include:
- Data format: Data coming from different systems, in different formats, all needs to be integrated and consolidated into the format your platform needs. Every customer is going to have their own rules they need implementing, whether it’s making sure First Name and Last Name are split into two fields, or converting multiple currencies to one.
- Data quality: Especially if companies are moving to your platform from an aging legacy system, there may be an expectation that ‘new system = new, improved data’, and that this migration will solve all their data quality issues. It’s on you to identify those data quality issues and clean up the data as it’s onboarded, not just propagate the same issues into a new platform.
- Not making your customer do the work: It’s common to make customers themselves do the work to get their data into the format you need. This can not only cause delays and bottlenecks (especially if the customer isn’t very tech-savvy or doesn’t have their own engineering team to help), but it’s not a great first impression when they’ve spent money with you. Wouldn’t it be better to make their life easier and take the work off their hands?
- Not overburdening your engineering team: You also want to make the data onboarding process as easy as possible for your teams. The goal here is to avoid using expensive engineering resources to custom build a data onboarding solution for each new client, but instead create a process that can be managed by less technical staff.
Automating the customer data onboarding process
The solution to all these problems is to automate the process. Building an automated customer data onboarding pipeline enables you to:
- Onboard more customers
- Do it more quickly and efficiently
- Do it without requiring expensive engineering resources
- Provide a better value proposition for your customers (both during the onboarding process but also in being able to continuously ingest live data easily).
Your automated pipeline not only needs to handle the entire end-to-end process, but it also needs to account for the variations in format, quality and frequency of data delivery that come from different customers.How Formula 3 Group were able to onboard more customers and grow their business without increasing headcount
Steps of a data onboarding pipeline
If we break down the individual steps we want a data onboarding pipeline to handle, they’re generally common to every job:
- Ingest: Taking data and ingesting it from a source system – whether that’s an API to an enterprise application, a database, a file dropped into an SFTP site, or anywhere else. The data ingestion process includes being able to handle whatever structure those endpoints might provide, even if that structure might be dynamic. We also need to account for situations where the source endpoints are unavailable or uncooperative.
- Transform: Data ingestion is usually followed by a transform phase, where you need to apply some rules to shape the data to the format you need. This stage can be where some more basic data onboarders fall down, with limits to how data is mapped. The advantage of a more powerful platform (such as CloverDX) is that you have complete control over what the mapping looks like, with the ability to build complex business rules to transform the data however you need as it’s mapped.
- Validate: Your pipeline then needs to examine the records in the dataset to make sure they meet certain rules before you proceed any further. This generally involves doing some data quality checks, applying specific business rules - both your business rules and rules specific to each of your customers - and pulling out records that are rejected and doing something with them, before you move on to the last phase…
- Deliver: The final step is to deliver the resulting data set to some target endpoint – a storage system, API, etc. This step not only includes the actual delivery of the data, but also monitoring to make sure it completes, and audit to record the fact.
Specifically for customer data onboarding, those broad stages can be broken down into specific steps that we want our automated framework to handle automatically:
- Detecting arrival of client files to be onboarded
- Detecting format and layout of client files
- Reading client files
- Assessing quality
- Loading to target
- Detecting/logging at every step
The steps of a data onboarding pipeline
The second part of this post looks at how you can automate this customer data onboarding process in CloverDX.
The CloverDX Data Management Platform is designed to build and operate data pipelines. It enables you to design pipelines in a visual editor, but also to code whenever you need, so you’re not constrained and can build a completely custom onboarding framework for your specific needs.
CloverDX can connect to any type of data, ingest it, shape it, cleanse it and write it to any target. And automation means onboarding jobs can run automatically and unattended, with monitoring and error alerting to notify you of any issues.
The video on the next post shows you the step by step process of how this works, and walks through how you can build a single pipeline to work with many different clients, by using configuration files to drive the whole process. (What does that mean? Mainly it means you don’t have to build a new pipeline for every new client, or for every change you need, and it means that this configuration – which can just be a human-readable Excel file – can be managed by less technical people).
- Read the second part of this post: Automating customer data onboarding: How to build an end-to-end pipeline in CloverDX
- Watch the webinar these posts are based on: How to build an automated customer data onboarding pipeline in CloverDX
- Read a case study of one InsurTech company who freed up engineer time by up to a third by automating their customer data onboarding