I worked for LiveRamp this summer in San Francisco and a lot of people have asked me what it does. While I know they’re working to resolve this problem, as of today, LiveRamp.com is sort of cryptic. It tosses around some words related to data and onboarding and partners and advertising, but it’s kind of unclear what they actually do.
In this post I will try to simplistically explain what LiveRamp does.
Before I begin, I do want to note that this does not represent the company in any official capacity, these are my understandings (some of which may be mistakes):
A long long time ago in an interweb far far away, websites that had a lot of traffic (let’s say CNN.com) wanted to display advertisements on their frontpage. They could choose to call up a specific company (let’s say Coca-Cola) and ask them if they’re willing to pay $x dollars for 1,000 impressions, a fancy word for pageviews. This problem grows like O(n) (“linearly” for those non-cs people) and while that’s not bad for a computer, when it comes to the real world, O(n) means unscalable. As CNN gets more users on their website they have to mantain these billing accounts, hire more salespeople, mantain relationships etc., On Harvard’s campus, The Crimson still asks their new members to sell ads this way.
Then came a revolution. Instead of selling to specific companies, CNN can instead sell banner space for an entire time period (e.g. one whole month) to ad-network middle-people. So instead of CNN populating that banner space with a Coca-Cola ad, that banner space would then just make a connection with the ad-network and the ad-network would populate that banner space with an ad. This is now an O(1) operation for CNN, which means it does not grow as CNN grows bigger. From the perspective of Coca-Cola, instead of having to talk to CNN directly they tell their ad-network how much they’re willing to spend, what kinds of people and online users they want to advertise to, and then their ad-network will deal with it. It does get much more complicated, if you want to learn about some of it, here’s a video: https://www.youtube.com/watch?v=1C0n_9DOlwE
Today there are these ad exchanges which allow you to (in straight-up realtime, 500ms, hire a gamer with fast fingers type-of-situation) bid on how much you’re willing to pay to display an advertisement (kidding about the gamer, it’s all automated obviously).
Note: this was started by (I believe) Double-Click which is now a subsidiary of Google.
Now during this process the ad-network can collect a lot of information about users that are seeing their advertisements. Because CNN is literally just sending a redirect to an ad-network, the ad-network can see the same user on a ton of different referrer’s sites: CNN.com, NYTimes.com, and (perhaps not as likely for someone who visited the previous two) Fox.com. In the redirect from CNN, CNN can also tell the ad-network the name of the article the user is reading, what articles the user read in the last 24 hours etc., etc., and based on that information different companies will make a different bid. Here’s an example:
Case 1: I visit CNN.com, the banner is an iframe that requests http://SomeAdNetwork.com?title=july_4th_is_coming and Coca-Cola bids more than Toyota, and a Coca-Cola ad is shown.
Case 2: I visit CNN.com, the banner is an iframe that requests http://SomeAdNetwork.com?title=how_to_buy_a_car and Toyota bids more than Coca-Cola, and a Toyota ad is shown.
In both cases, the companies have more effective ads and the user sees something that’s relevant to them. All is well if you can get over the fact that this is the equivalent of looking at a billboard on a highway after having the billboard look at you and deciding what to show.
An issue arises though. Let’s say you’re Toyota and you have a list of customers that have their car lease ending in 2 months. You would love to find these people and display an ad to these people online. The old method of doing it would be to
A) find some identifiable trait of these people (e.g. more people in Duluth, GA or more people who read articles about monkeys) and
B) get around a .000000000000000…000001% hit rate on your target by hitting those identifiable traits. All-in-all; this sucks!
This is where LiveRamp comes in (phew…finally…). LiveRamp will help you find those people online! Woohoo. That is all. Thanks for reading. (Haha, just kidding).
So while this is what LiveRamp does, there’s some nuances that have to be deal with.
Let’s begin by how they may solve this issue. LiveRamp has two sources of data:
1) Linkages between emails, postal addresses, names, phone numbers etc., from straight up data companies with which they construct a massive identity graph.
2) To tie those identities to online entities, they essentially pay for web-traffic. So let’s say when you log into RandomWebsite.com with an email address, there’s literally a 1-pixel, invisible image that redirects to liveramp.com with your hashed email address. Then LiveRamp can attach a cookie to your browser and match you with an entity in their graph.
Now when Toyota uploads their file of postal addresses of people who’s leases are coming up, there’s LiveRamp essentially knows the cookie-id of those people.
Then LiveRamp distributes this data to the ad-network through a few methods. One example easily illustrated is called a client-side sync. Which means the next time LiveRamp sees you online they will redirect your traffic with some data-segments attached to your ad-network. For example:
Step A: You login to RandomWebsite.com with email (RandomEmail@gmail.com)
Step B: RandomWebsite.com embeds an image which goes to LiveRamp.com?email=[hashed version of RandomEmail@gmail.com]
Step C: LiveRamp has seen you before and has a cookie on your browser so they redirect you to AdNetwork.com?data=TOYOTA_LEASE_EXPIRING
And voila! The ad-network now can attach their own cookie to the browser and knows that the browser is one in which a Toyota lease is expiring. Within 5-7 days of uploading their file, Toyota can configure with their ad-network how much they’re willing to bid for someone who’s lease is expiring (along with all the old information about what their reading etc.,). This is what LiveRamp calls “data-onboarding.” Customers like Toyota can take their offline data and onboard it to their online advertisers through LiveRamp! 🙂
I do want to note that LiveRamp does a ton of work to ensure privacy. There used to be a company called Rapleaf which is very closely related to LiveRamp which did have some privacy issues. A lot has changed since those days. For example, PII (personally identifiable information) is never stored with online information. A single company’s data is siloed specifically for that company and not given to anyone else. There are a TON of other privacy precautions that they take and I’d say they’re one of the leaders in ensuring onlining privacy. You can, for example, easily opt out of LiveRamp by seraching for “liveramp opt out” in Google.
This is what LiveRamp does and it’s pretty sweet. They claim to get around 2 billion requests a day and with roughly 30 engineers, have a higher data to engineer ratio than either Google or Facebook. It’s tons of really smart people with a lot of grit and a lot of heart who are working on really cool, fast-paced stuff.
The CEO, Auren Hoffman, once told me something along the lines of: “I like really love my family, but the next group of people I really love is LiveRamp. I expect like 98% of people here will one day start their own companies and like, I’m not even going to ask what they’re doing, I’ll put money in. We’ve spent tons of resources in growing people and I think we do quite a good job.”
A huge part of their hiring process is looking for great communicators and teammates and it works out! They’re growing like crazy in the next few months…so if you wanna, here ya go. :).
To find out about what I did specifically during my internship, do check out my next post!.