Module 3 - Different types of audiences and audience evolution

How do you build the perfect audience?

You do not build it from a guess or a personal preference, you build it from conversion data, through a test, learn and implement strategy, and it takes time to understand where your conversions are actually coming from.

This is one of the most common questions asked, and within video campaigns you have the option to run affinity, custom, demographic, in-market, remarketing, customer match and similar audiences. You cannot run all of these at once unless you have an enormous budget, because that would mean testing twenty, thirty or forty different audiences in a single account.

What are affinity audiences?

Affinity audiences reach people based on their interests, what they do and their habits.

That is a good starting point for understanding who you might want to reach, before you know anything more specific about their buying behaviour.

What are custom audiences and when should you build them?

Custom audiences let you reach people based on what they are passionate about, their habits and interests, and their purchase intent, but they are not usually where you start.

We use custom audiences a great deal, and they act as a bridge between affinity and in-market audiences. Once you have learned from affinity and in-market data, that is when you move on to building custom audiences based on what you now know about the people responding. Step one is rarely custom audiences on day one, you start with affinity and in-market, overlaid with demographics, and once you are seeing a level of success you move into remarketing and customer match. Once those are succeeding too, you move into similar audiences. It is a flow: start with Google's off-the-shelf audiences, work out where your success is coming from and why, then build custom audiences, which is where your ultimate success lies and where you really start to understand how people are interacting with your product or business.

What makes an in-market audience so useful?

In-market audiences let you overlay who someone is with what they are actively trying to buy right now, which gives you a real advantage over platforms like Facebook.

I really like in-market audiences, because Google is looking at people who are actively searching for a particular product. Overlay that with affinity and you could find, for example, golfers who are currently in the market for a golf product. That combination of who someone is with what they currently want to buy is where the advantage over Facebook shows up. Facebook can say I am a golfer, but as it happens I have not picked up a golf club in twelve months, so I am not in the market for golf products, I am just still watching golf on television, following the PGA Championship. Facebook, and probably Sky and other TV channels, would still target me with golf products because I follow the sport, even though I am not currently buying. An in-market audience only catches me once I am genuinely in a buying mode, say if I suddenly start visiting golf websites or looking at golf products on Amazon.

How far can you take this with custom audiences?

You can build custom audiences directly around competitors, including their websites, their checkout pages, and even their membership areas.

That lets you reach people who are actively buying, actively in the market, and closer to the perfect customer, because once you know who someone is, what they like, where they have been, and that they are in the market, combined with a basic demographic overlay, you can put an ad in front of somebody at exactly the right time. That goes well beyond what most people are doing on Facebook, reaching someone who is actually ready to buy a particular product.

Where does Facebook get its in-market signals from, and why has that changed?

Largely from behavioural signals like what you talk about on WhatsApp, and Apple's iOS 14 privacy changes have made that harder for Facebook to rely on.

If I sit here and talk a lot about a particular type of wallpaper or a particular type of car, I will most likely start seeing ads for that product in my Facebook feed, which is how Facebook has been building its in-market data. Since iOS 14, the privacy changes Apple has layered on top of that have made it a lot more difficult for Facebook to do this, and it will keep evolving over the next couple of years. Google, meanwhile, is moving more towards what it calls federated cohorts, categorising people in-market based on the websites they visit, their behaviour, their age, household income, family makeup, and the other products they search for.

How personalised can Google actually make an ad?

Personalised enough to change not just whether you see an ad, but which version of the offer you see, based on your search and browsing behaviour.

One example I give is searching for an Italian restaurant in Manchester, which is about fifty miles from where I am in Doncaster. A standard search for Italian restaurant Manchester gives you a generic list, but what you actually want is the perfect restaurant for you. There are hundreds of Italian restaurants in Manchester, and if I do not mind paying a premium, want some entertainment, and do not want a chain restaurant, that should change the parameters of what gets shown to me as an ad. Google understands this from search behaviour, how you engage with other websites, and the amount of time you spend on those sites and on videos. What we are looking for throughout this course is engagement, and the perfect audience, not a spray and pray approach that just targets everyone searching for Italian restaurants in Manchester regardless of what they actually want. We would love to see the inner workings of exactly how the algorithm decides all of this, but this is the way to think about audiences.

What is the right order to build your audience structure?

Start with affinity and in-market, then build custom audiences, then remarket to your successful conversion events, and only then spin out similar audiences from what has already worked.

In terms of structure for your YouTube ads, we start with the opportunities in affinity and in-market, then build custom audiences, then remarket once we are seeing success against our conversion events, whether that is people spending more time on site, watching more of our videos, or generating more sales, conversions or enquiries. Only then do we spin out similar audiences from our successful ones. Do not start building similar audiences on day one: they come from things like people who have been on your checkout page, made a phone call, filled in a form, spent a certain amount of time on your site, or watched three quarters of your videos. Build your conversion events with that longer term development in mind, because it is the continual refinement of these audiences that lets you build the perfect audience over time. Hopefully that gives some insight into where we are trying to go, how the structure of a campaign works, and the order in which we move from what Google suggests through to designing and optimising our own audiences.