Data Operations: the Fuel for Your Business
At the intersection of your system stack lies opportunity
What is Data Operations, Anyway?
Data Operations is a hidden team in your organization, connecting business and people systems and helping Sales Operations, Marketing, and Customer teams to present accurate metrics, marketing data, and customer operations to executive leaders. You might not realize you need a Data Ops team until you find unexpected results in data that slow down the business. And if you had a team like that, you might find unexpected advantages too.
But why Data Ops instead of just … Operations? After all, the people who solve Operations problems probably know about the data in those systems, right? Too often operations teams get siloed in a specialized system and don’t have the overall picture of how data interacts in a complex system caused by the interaction of multiple systems and teams.
Data Operations might refer to the people who perform data tasks (usually in a Sales Ops or Marketing Ops team) or might be a dedicated analyst or team working on Business Intelligence analytics. It covers the following:
Data that is shared between systems
People who analyze, update, and fix the data in those systems
Tools they use to keep their data accurate, relevant, and timely for business
And the governance process they use to keep everything working
This post suggests a new idea. What if the Data Operations function were elevated within every Go To Market (GTM) organization or Revenue Operations team as a key tool to increase sales velocity, pipeline growth, and customer happiness? This is beginning of that story. Let’s Go!
Needed: a new way to talk about data
We talk frequently about the need for better data but not always about what that actually means. Data is the connective tissue between the different parts of our business. When it doesn’t work or causes problems it creates hidden and visible problems that prevent businesses from running as expected.
Yet little effort is spent to fix data on an ongoing basis. Everything is a one-time fix, prompted only by a problem in the business. What if the problem is that we’re not thinking about data operations as an integral part of our business and instead as a bolt-on solution? Rethinking the health of data in your business is an important task. Heeding the adage of “garbage-in, garbage out” is a good first step and there’s more to be done.
Why would you see these issues as systemic and not just as one-off problems? Here are a few examples. As a developer, I’ve built applications that depended on data. As a team or business leader I’ve used both internal data and external enrichment data for customer ops, marketing ops, and sales ops tasks. As a subject matter expert in Data Quality, I’m responsible for data that drives the business and accountable to other teams to fix problems that emerge in our data.
A basic start to fix these problems involves quantifying the metrics and measures that control data. It also involves building dashboards and reports to highlight problems and successes that might not be seen immediately. You might also have reacted to seeing data that spans systems by trusting certain kinds of data too much, and tried to do things with superhuman personal effort instead of building automated systems.
It’s ok! You can do better by making different mistakes (try not to make the same ones repeatedly) and putting techniques together to provide a “data weather report” for your organization. If you’re doing something daily or weekly, it can probably be automated and improved.
This article – the beginning of a blueprint for a Data Operations practice – will not solve all of your data problems. It will also not tell you exactly how to untangle the particular challenge you’re having with your business.
It will illustrate a different way of thinking about information to inform inputs to your business challenges that depend on data. It will help you to frame the data challenges you see in your company that today look like non-compliance or general error. And it will help you to become literate in the data of your business and how it is interconnected across organizations.
By building a Data Operations practice, you will better understand how your company lives, breathes, consumes, and exhausts data and reveal key information and insights.
What kind of data exists today
We’re creating more and more data all of the time. This should be obvious even if you haven’t taken the time to think about it lately.
The website HostingTribunal.com has gathered some relevant stats to quantify the amount of data we’re creating, and I’ve shared that information below:
Looking at the aggregate data presented above there is no doubt that we are swimming in information. No worries, you say - the consumer market is much different than the enterprise market - and the data that you see there is not really representative of what we need to see in our businesses.
Yet we all use consumer devices (phones, laptops) and our attention span is limited. Our calendars, messaging requests from slack, and general information load is making it hard for people to keep up.
Cognitive load is a real problem. We use tools like checklists to help us process information in the right order, or personal assistants like Siri and Alexa. But there remains a gap in the understanding that we have when looking at the data itself: how it all fits together.
Consider a simple example: the account name of a customer. When you refer to the “name” of that company you might be be referring to some of the following possibilities:
The “simple name” of the company, like what you would use in an email. A local hardware store might be colloquially called “Duvall Hardware”, even if it belongs to a chain like True Value
The legal name of the company - this might be different from the name of the company or the franchise
An advertising name, like “Duvall True Value”
When you find any of these names in your sales, marketing, or financial systems, how do you ensure they point to the same company who is your supplier or customer?
And how do you make it clear from system to system that the company referenced in your sales system is the same as the company with the same name in your marketing system?
Add to this conundrum the fact that you have lots of people in your organization entering new information all of the time and you start to get the idea of the scope of the problem.
Data - or the information you use to run your business systems - is everywhere. When we talk about this information, what do we mean?
Data Operations is not only the data we use but the way we use it
For the purpose of data operations, we are referring to everything we put through our operational systems that helps us run our business.
The purpose of a data ops team, and an operational team in general, is to make your business run more smoothly and efficiently according to the business rules agreed to by stakeholders. When the business cannot run well, a data operations team might suggest changes (start doing something, stop doing something, or change an existing process) to execute and measure with the goal of helping the business.
Think of data ops as a consultant to identify the changes that are shared between systems, not just the changes in any one system. For each team in the business, the proposed changes that need to be made to improve a process or fix a data problem will be different.
The place where that change happens will also be different. It is critical that each team maintain a system of record that is the undisputed source of information shared with other teams so that there is no conflict or confusion when finding the details that team uses to run their business.
For a sales team, that system of record is often their customer relationship management system, perhaps using Salesforce to store this data. Using Salesforce as the system of record gives the rest of the company a place to review information about ongoing sales business and a method of interconnection from that sales data to other systems that store information relating to companies, people, and customers.
A customer success team might store different information about customers (and prospects), focusing instead on the interactions they have with those customers. When a customer contacts us, we might want to know how many times they have contacted us previously, what entitlements they have in their contract, or what products and services they have bought. Ideally, we want to be able to link the people contacting us now to the actual customers we have in our database, without duplicating those company links.
A marketing team might store still more information about the initial contacting of customers, linking them to a campaign or a lead gleaned from a particular source or event. For the marketing team, attributing the initial interest from a lead to the end product of a successful sales activated lead or a sale is the key to understanding the input costs of the business. Should you spend money on an event or a campaign? Getting the results and tying them to the business sales is a key way you would do that.
Now, think about these three hypothetical teams and imagine how they want to work together. In the perfect scenario, they all have equal access to information about the customer, suffer no friction when trying to identify a person who has previously or newly contacted the company, and have a direct line to quantify the cost of revenue to the leadership of the business.
Have you ever worked in a business that had all of those things figured out? Maybe. But probably you are familiar with some of the pitfalls that occur when the data machine is not quite so finely tuned as in our hypothetical example.
Pick an item to dissect and you will find a pattern in that same hypothetical pointing back to an issue in the Data Operations of that business.
Take account names again as the example. When you encounter a multinational corporation that uses the same name in multiple territories, how do you uniquely identify each contact and each company so that you associate them to the right company?
In the perfect scenario you might have a global account model and the ability to map the person who contacts you to the office where they work. In reality, you’re probably going to name the company the same and use a different field somewhere in your system to indicate country, ownership, or a legal name that allows you to separate an office for a company in New York and an office for that same company in London, Bangalore, or Tel Aviv.
Perfect scenarios do not exist in a world where data is entering our systems from everywhere and not always in the right format or type. So how do we build a structure, a team, and a process for improving Data Operations in a business not just one time, but on an ongoing basis? And how do we do that across systems and business units? We want to end up with a situation where people no longer talk about how data quality is broken and start talking about the different and new things they can consider with integrated, updated, validated, and trusted data.
Why should you care about having good data?
According to ZoomInfo, approximately 40% of leads contain inaccurate data, and almost 30% of people change jobs annually. It’s easy to see the impact of having bad data when you are trying to contact prospects, and this extends to the way you use information in your business, not just with customers. In 2019, 64% of marketers reported that improving data quality was their most important challenge. It’s clear that when you use data well you will help more than one department at a time.
Better data is a force multiplier that helps multiple areas of your business improve at the same time. In 2015, Sirius Decisions estimated that an organization with strong data practices could generate up to 70% more revenue than one that did not have these measures in place. We focus mostly on the revenue generating aspect of having good data and don’t always think about the wasted work that occurs inside of organizations to fix data shared between departments.
In short, fixing your data will help your business be more effective.
Leading businesses take these basic practices and build a machine to improve their information on a continuous basis to provide a self-healing system even when garbage data gets into your system.
Ok, Now What?
You want to fix your data, and you realize you might have more than one problem you need to tackle. Don’t try to boil the ocean. A good first step would be to find one area of your business that needs improvement on a one time basis, and to think about how you would improve that data on an ongoing basis whenever you receive new information of that type.
Let’s say you wanted to improve the accounts in Salesforce, and you knew that one area that always seems to be wrong is the web site field.
A blueprint for fixing this might look like a typical product management story:
“As a [stakeholder of the data] who needs to use [type of data] in [system], I need to know that the [object or field] is always following these rules [a list of rules]”
Translated, that might look like:
“As an Account Executive who needs to use Accounts in Salesforce, I need to know that the website field is always following these rules (is not empty, has “https://” at the beginning of the website domain, and has been checked within the last year to be an accurate web page for that account)”
Applying this through a Data Operations lens might include:
A list of the stories you want to approach, and a goal for when you want to release that functionality
A testing plan for known accounts and new ones that have a website
A method to assess existing accounts for websites that don’t meet this criteria
An enrichment capability (manual or otherwise) along with a process to test whether the website connected with an account is the right one
A simple report or dashboard indicating the proportion of accounts that don’t meet the criteria
As you get more comfortable with this idea, you might extend it to:
Examine other systems where this type of data exists
What to do when you encounter a problem and how to remediate it
If there is a dedicated process that imports this kind of data, how could you incorporate this change into the dedicated process?
Fixing your data won’t happen overnight, and data doesn’t always stay fixed. Applying a data operations mindset to the people, process, and tools in your organization will help the data that exists between systems to stay in sync.