When do you need to make an ID Graph?

Alright folks, it’s been a brief minute, but I’m back and, more importantly, I’m going to dive deeper into the topic of one of my favorite recent podcast episodes. A little while back, I chatted with my friend Ray Esteves of True Influence on the Corporate Data Show. If you’re a listener, you may recall that we were originally going to discuss machine learning, but decided that was a snorefest (I still stand by that) and went with identity graphs instead.

Like I said, that episode came out a good minute ago, so even the most dedicated CDS listeners might need a refresher on what we covered. Luckily, I’m going to do much more than a refresher, so sit back and read on to learn everything you need to know about ID graphs and their usefulness (or lack thereof) to your organization. So, let's begin the discussion on The Great Identity (Graph) Crisis. 

What is an ID Graph?

An identity graph is essentially a database that both compiles and displays all pieces of information that correlate to individual customers. This is done through the process of identity resolution, which is the practice of piecing together as much info as possible, usually from many miscellaneous sources, to create an accurate, living profile for an individual customer. 

However, to be particularly pedantic, when most people say “individual customer” in this context, what they’re really talking about is a household (or a business, account, department, etc.). This is most accurate because in many cases, multiple people in a household or other organization are using the same accounts, and these accounts provide a lot of the data in your graph. So, keep that in mind whenever I refer to customers moving forward.

You might be thinking that this sounds a lot like a CRM, and I get that. They’re both methods of storing large amounts of customer information. However, an ID graph is even more comprehensive, as CRMs are just one of the many places they pull data from. Most modern companies collect and keep data in numerous places, including CRMs, email databases, eCommerce softwares, social media analytics, and many more. ID graphs are used to identify any data across all these sources that correlate to the same customer, and triangulate it into one database.

How accurate are ID Graphs?

For the most part, they’re extremely accurate, but it does depend on which kind of matching is at play: deterministic or probabilistic.

Deterministic matching takes place when a customer’s data is matched with 100% certainty. This usually means that a concrete identifier is being used, such as a personal email address, username, or phone number. As you can imagine, when deterministic matching is the primary method through which your ID graph’s data is triangulated, it will be at its most accurate and reliable.

As you probably already derived from the name, probabilistic matching is the practice of making extremely educated guesses based on less concrete identifiers like IP addresses, operating systems, etc. For example, if the same website is being regularly accessed from the same device at the same IP address, it’s safe to assume that this is being done by one customer, or at least by members of the same household. So, while probabilistic matching may not be 100% accurate, it still can and will provide a solid ID graph for you to make educated decisions.

Okay, so how does all of this help my business?

Identity graphs benefit you and your customers in numerous ways, but the most obvious is in how feasible they make personalization. As a customer yourself, you’ve probably reaped this benefit hundreds of times without even realizing it. Every time Netflix recommends the perfect movie, or Instagram shows you a post that inspires your next outfit, there’s an ID graph working behind the scenes, tracking your online activity and attaching it to a living profile of you that allows businesses to cater to you. A good ID graph can provide such specific info, in fact, that it can guide businesses on when to advertise to you, where you’ll be, and even why.

For example, let’s say you live in the suburbs but work in the city. You only have one car, and your spouse needs that to get to their job, so you’re stuck taking the train in every morning and back out every night. You’ve had another auto purchase in the back of your mind for awhile, and have searched local listings, but haven’t pulled the trigger yet. Well, every time you searched “2021 Toyota Prius,” your browser tracked that. Plus, your location data shows exactly when you’re on your commute every morning and night. So, just as you’re about to lose it and tell off the guy manspreading next to you on a packed train car, your Instagram feed will show you an ad for the local Toyota dealership, with a special offer for a year of 0% financing. And let’s be honest; you’re going to click on it. That perfect targeting was made possible by nothing other than an ID graph.

That example is, of course, more focused on the B2C side of things, but personalization comes in very handy on the B2B side as well. Any time a salesperson or other agent for your company contacts a potential client, they can have that person or organization’s profile pulled up right in front of them. Just like that, they have easy access to their company and position info, purchase history, and even little things like a certain client’s marital status, or the names of their kids. This allows them to be more human when making deals, because they can show that they care. It’s like I always say, to the point you’re probably tired of hearing it: automate what you can, so you have time to be human when it counts.

Do you need to build your own identity graph?

In short: no, probably not. The amount of data necessary to justify building an ID graph is much larger than most companies are working with. For example, over at True Influence, Ray Esteves is working with upwards of 160 million contacts. So, it’s no surprise he needs some help keeping them straight. 

He says — and I agree — that to have the amount of contacts needed to effectively triangulate data, you need tens of millions when dealing with B2B, and upwards of 300 million for B2C. And quite frankly, even if you have that much data sitting around on some server, it’s a huge commitment to build your own ID graph. It would take a lot of time and effort, so much so that it would probably make more sense to just hire someone else. So, the harsh reality is: if you want to get a good ID graph, you’re going to end up paying someone for it.

Using third-party data to bolster your id graph

That’s the first scenario in which you’d have to pay someone for an ID graph. The second is when you don’t have that much data on hand to begin with, which is most likely the case, and you need to buy some (or a whole lot). This will give you the resources you need to effectively triangulate data between your incomplete datasets and other data that’s floating around out there. Then, you’ll finally be able to reap the aforementioned benefits of a solid ID graph. 

There are lots of companies out there that compile and resell massive amounts of data that can be used to bolster your graph. I can’t really think of any right now, but if I do, I’ll be sure to direct you to our websi– I mean, I’ll be sure to let you know.

Get ouT there and get graphing!

Like I said, if you’re a small business, an ID graph is probably not a worthwhile investment. You’re not working with enough clients to need help keeping them straight, and you don’t have enough contacts to triangulate data anyway. However, if you’re right there on the border of scaling your business and really taking off, a great ID graph just might be the thing that gets you there.

In the meantime, be sure to check back in for more tips to help you do your best work, and hit us up with any and all of your data needs!