Here’s How Advertisers Use Data To Target You With Ads

It’s remarkable to believe this John Wanamaker phrase dates back to over a century ago – especially when you consider that it’s only been in the last decade that we’ve had any knack for resolving ad spend dilemmas.

This modern breakthrough can be pinned back to using higher-fidelity data sources and by better understanding user signals (produced both online and offline).

So, let’s chat about data.

Historically, advertisers largely bought audiences using a scattergun approach, like purchasing TV spots during times and on channels commonly watched by the advertiser’s target demographics. But in the past decade our ability to collect and understand a growing amount of data about user behaviour has enabled us to move away from a ‘spray and pray’ approach.

Advertisers now buy using an audience, mostly programmatically or via a data-informed walled garden like Facebook or Google.

Most often, cookies (pieces of code dropped by a website on a user’s browser to track their behaviour on that browser) or device IDs (numbers associated to your mobile, iPad or PC) are the main signals that we use to inform these audiences.

But this is a fairly flawed system, as cookies don’t necessarily represent the same person. Each individual can have dozens of cookies and device IDs. Think about it – how many people use any combination of phone, tablet, laptop, desktop and smart TV every single day? 

As advertisers in Australia, we’re starting to move beyond cookie targeting, with better signals used to inform audiences.

We can break down data used by advertisers into 3 categories:

1. Panel data: iCCS and inferred data used for insights, trends and to plan audiences
2. Device data: Cookies and mobile devices
3. Person data: Personal information (PI) such as names, email addresses and phone numbers

These three types of data can then be described in one of two ways – inferred or deterministic.

Inferred/Probabilistic (panel/device)

Panel and device data are forms of ‘inferred’ or ‘probabilistic’ data. These are essentially data points and characteristics assigned to a person based on their activities and behaviours online.

An example would be assuming that devices in the same geo-mapped location (like the address 123 Fake Street) belong to the same person. This isn’t known through personally identifiable information like logins – it’s just inferred or predicted.

Deterministic (person)

What we’re getting to now is people-based marketing, which relies on deterministic data. We’re not making any assumptions or inferring anything about these users. The data set is 100% accurate (more or less).

It’s associated to individuals using different methods like logins, purchases, social platforms and authenticated services – basically something enabling us to attach information to an individual.

That’s all well and good, but how do I collect all that juicy data?

Thankfully, this is probably the section you already know a little about. Our favourite three parties – the first, second and third. These also correlate fairly directly to the order of their fidelity – being the data set’s integrity and relevance.

First Party

When talking about our advertisers’ own customer/website data, this is first-party data. It’s data owned by you. For example, we may drop a pixel onto BMW’s site to collect data about how users are engaging onsite. From there, we can segment these users and build audiences. Using first-party data, we’d be able to target users who have researched the M140i but haven’t enquired yet.

Second Party

The easiest way to think about second-party data is imagining other people’s first-party data. Businesses can use this to enrich their own data sets. For example, News Corp and RealEstate.com.au could make a deal to enrich each other’s understanding of their combined audiences by sharing their first-party data. Second-party data typically has the highest integrity after first-party sources.

Third Party

Third-party data has the least fidelity but also the greatest scale. This is data aggregated or compiled by a data vendor. You have no control over how they’ve put these segments together – as they’re simply providers trying to drive scale using their own first- and third-party data. For example, you might target a third-party audience of users who have a household income of $200K with no kids based off census data.

Where are we moving with targeting?

As advertisers, our end game is to understand each individual on a person-by-person basis and to accurately reach users when they’re most receptive to relevant advertising. Unlike the US (who have played in this space for two decades), Australia is facing some barriers in this shift to people-based marketing because it relies on multiple publishers sharing data between themselves.

And that’s definitely a big ask when data is power.

Hungry for more on data? Get in touch with one of our experts in Data & Analytics today.