Digital targeting demystified
This is an ambitious post because it is my goal to finally end the confusion surrounding how to target audiences with digital media.
Let me start by clearly stating that deciding which segment to target should be done further upstream as a part of your strategic work.
By the time you reach the pointy end of the marketing process (digital targeting), you should be super clear about who you are trying to reach, and what your positioning is for that audience. If any of this sounds new to you, I highly recommend the marketing mini-MBA taught by Mark Ritson for getting your marketing basics sorted.
In a hurry? Jump straight to these sections.
- Digital targeting 101
- Demo targeting
- Profile targeting
- Contextual targeting
- Behavioural targeting
- Lookalike targeting
- Customer targeting
- Geo targeting
- Device targeting
- Placement targeting
- Campaign objective, it matters
Why is there confusion?
My working theory is that the confusion with digital targeting stems from a few different factors:
- Digital is still relatively new, although Google has been around for 20 years (wow!).
- There are a lot of metrics.
- There are a lot of options to "target".
- Things change very quickly. New features are released all the time.
- Ad sales teams push products. Betas anyone?
- Folks tend to believe that digital is a silver bullet. It is not. See above marketing 101.
- FOMO is a real thing.
As someone who drank the Kool-Aid, trust me when I say this:
It is very hard, if not impossible to create long-term strategic advantages with digital media targeting. The gains are short-term and marginal because the tools provided by Google, Facebook et al are available to all advertisers. It's only a matter of time before someone else catches up to you.
What does that mean for you
Yes, you should make sure that you are executing well. No doubt about that. And digital targeting is an important part of good execution. But it should not be a distraction.
I want you to
- Understand how this targeting business works.
- Build your targeting playbook.
- Move forward.
Occasionaly you may need to update your targeting playbook when new things emerge. But there's a good chance that it will be a version of something that already exists.
Ok. Let's dive into the meat and potatoes.
Digital targeting 101
- If it's broader it tends to be cheaper. And the corollary to that statement. If it's narrower, it tends to be more expensive.
- Targeting is actually fuzzy, especially when it comes to behaviours.
- If it sounds smart or sophisticated, ask how the sausage is made. 9 out of 10 times it is not as fancy as it sounds.
- The big platforms have a lot of data on you. Google, Meta (formerly the artist known as Facebook), Amazon, LinkedIn, and now TikTok. The more data a media vendor has, the more reliable their targeting tends to be.
- Your own data can be valuable. Use with care.
- Always ask yourself, should you even be doing this? See above, marketing 101.
My taxonomy for digital targeting
Platforms tend to call the same thing by different names so I built a taxonomy for classifying all digital targeting.
Targeting based on age and gender criteria.
It is the simplest kind of audience targeting available, but, in my view, can be really powerful to reach a broad audience (see above, broad tends to be cheaper).
Platforms that require users to log in (and declare their age and gender) have better quality data on demographics. This is true of social media platforms like Facebook, Instagram and TikTok. Google doesn't require you to login into search, but they nudge you to do so in many places (YouTube, Chrome) or make you (Gmail). Their machine learning algorithms are powerful enough to estimate your demo and age based on your behaviours and partially declared information.
The big tech platforms can do a pretty job of demo targeting due to their large user base, and access to more data, which helps makes their algorithms smarter. Big tech = better targeting options. This will be a recurring theme as we explore other targeting options.
Age and demo targeting becomes a bit fuzzier on the open web.
Since we don't need to identify ourselves in many places on the open web, companies that run targeted advertising in these environments rely on models to estimate your age and demo. Don't get me wrong, these models do a pretty good job of estimating who you are but there's a natural limit to their accuracy. They can never be as accurate as targeting within environments where the user has declared their identity.
Demo targeting is the bread and butter of digital targeting. Use it when you want to reach your total market.
Targeting based on a user's profile.
The only platform I feel is worth mentioning in this segment is LinkedIn. Your profile on LinkedIn tends to be more accurate and comprehensive than anywhere else on the internet. It is your digital resume after all. LinkedIn has lots of data on your professional identity which can be used to target based on a variety of criteria like seniority, company, function etc.
When your profile is comprehensive and accurate, this is the good stuff.
Targeting audiences based on the content they are reading or watching.
If someone is reviewing different SUVs on a website, there's a good chance they might be in the market for a car, maybe even an SUV. Not a bad way to target an audience right? Right.
Reading about high interval training, you might a good candidate for workout supplements or even meal plans.
Again, this is the simple stuff. But the good stuff. Contextual targeting, content targeting, and keyword targeting, all do very similar things. They try to make sense of the content on a web page or video and use that context to target audiences.
Watching DIY videos on YouTube? You might be the perfect target for a home & garden brand. Ok, you get it.
I like contextual targeting because it does what it says. I can run my own tests and work out which contexts, websites and content generate the most sales or leads. Easy to explain to a CEO or CFO as well. Big plus.
Targeting based on user behaviour.
There are a lot of targeting options in this category. All of them classify a user into some behavioural cohort based on observed behaviours.
For instance, let us say you need to buy health insurance.
You may start by searching for this on Google. You will probably visit the website of several insurance providers and perhaps even visit comparison sites to compare different options.
Google can use these behaviours to classify you as someone in-market for an insurance product.
You may also over the same period visit a news website like The Guardian and read an article on the Great Barrier Reef. This might signal Google to classify you as someone who has an affinity for the environment.
A cosmetics company that has launched a new sunscreen product that claims to be friendly to marine life might want to target you using the affinity segment.
I personally stay away from affinity segments or interest-based segments. Firstly, these can be super fuzzy. The same person who cares about the environment can also have a love for sports cars. Yes, we humans are complex.
And secondly, I have found that a combination of demo targeting and setting the right objective (more on this later) does the job better.
Every platform has access to different data about you. The more of it they have, in terms of volume and variety, the better their behavioural targeting tends to be.
For example, Amazon has its own set of behavioural targeting segments which are based on your browsing behaviour on Amazon and its other properties. They can make light work of identifying someone as a new mum simply by observing that person buy nappies.
So, are cohorts with purchase intent more valuable than affinity or interest segments? It really depends. If your brand is relatively unknown, and you have a smaller budget than your competitors, these in-market or purchase cohorts can be really valuable to you. They give you a chance to get in front of customers who are going to buy soon. But does this mean, these are the only cohorts you should target? Again it depends.
With behavioural targeting, I generally recommend cohorts with purchase intent, but only in those environments where I believe purchase intent is clearly identifiable, i.e on Amazon or Google. There are other ways to identify intent, and that is to use lookalikes.
Targeting audiences who look like some audience.
Lookalikes, actalikes, similar audiences. They are all the same thing. What they have in common is that they need some seed audience for the platform to go and find others like them. I will use the term lookalikes to refer to this type of targeting going forward.
The seed audience can be created from
- A customer list (emails or phone numbers)
- Website or in-app behaviour
Why would you use this?
If you have a list of existing customers, you can use lookalike targeting to find more people like them.
- The quality of your seed audience has a positive correlation with the quality of your lookalikes. Garbage in, garbage out.
- You give up control to the platforms to find this audience. It's a black box. There's no guarantee these lookalikes will be any good.
- When the audience becomes large enough, lookalikes start to look like broad demo targeting. Go figure.
The hall of fame of what not to do with lookalikes:
- Lookalikes for every page of your site. Web page URLs can be decent proxies (e.g. add to cart, or visit specific pages) but if you build a lookalike for every page on your site, you end up with several audiences that look kind of the same. Pick certain key pages to build lookalikes if you must.
- The seed audience set is rubbish. E.g. list of emails from a database that is outdated and doesn't really reflect the customers you want to target. See above. Garbage in, garbage out.
- The seed audience is too small. When the seed audience is too small, the algorithm may not have enough signal to create decent lookalikes. I'll be cautious when this is the case.
I like lookalike targeting because it is relatively easy to fire up. I pay attention to the quality and size of the seed audience and use cautiously when either of these is no bueno. It can work wonders, especially on social media platforms where we don't have access to in-market segments off the shelf.
Targeting those who have visited your website, app, or engaged with your ad (some platforms allow the latter).
You need to have some way to build these audiences in the first place. You commonly use
- A platform tag
- An analytics tag (compatible with some platforms, not all)
Shameless plug - if you need an intro to tag management, my essential beginner's guide to google tag manager takes you through the basics of deploying marketing tags on your site.
The worst practices from lookalikes apply here as well.
- Don't make endlessly granular retargeting lists. Identify only those site triggers that are good proxies for user behaviour.
- The size of the list matters. If it's too small, it may not work.
- The age of the list matters. If it is outdated, it may not be worth using. All lists have a shelf life and should align with your typical customer journey.
Lots of people run retargeting ads. It can work, but I'm not entirely convinced it will drive incremental sales. Here's a radical thought. Try stopping it and see if it has any impact on your overall sales. Ask for forgiveness, not permission.
Targeting your own customers using a customer list.
I call this customer targeting, but this also includes targeting a list of leads or warm prospects. Basically targeting people from a list.
You will need to upload or connect your customer data to the platform you use for buying ads (Google Ads, Facebook etc.). The platform references a unique identifier, such as an email or a phone number from your list, to find your customers and target them.
- Platforms cannot match 100% of your audience. The match rate can be as low as 30%, or as high as 70%. For example, if your list contains your customers' work emails, it may not match their Facebook profiles where they might use their personal emails.
- The size of the list matters. If it's too small, it may not work. Platforms have thresholds on how big a list should be. A list of 20 is no bueno. Take them out to dinner instead.
- The quality of the list matters. Garbage in.. you know it by now.
It makes sense to target your customers sometimes. It makes more sense to exclude them many times. It can be useful. Or if you are a fan of Byron Sharp, just target the market (aka demo targeting) and be done with it.
In addition to the targeting methods covered in the previous section, we have a few more options to target in digital media. These can be used in combination with the targeting options previously covered.
Targeting an audience based on some geographic or location criteria.
Different platforms have different capabilities when it comes to geo-targeting.
- Target entire countries
- Target states and regions
- Target cities or towns
- Target postcodes
- Target suburbs
- Target a radius around an address
You can combine geo targeting with all the other types of targeting we discussed so far. I would be careful when doing this so that the size of audience doesn't become too narrow.
There may be internal business reasons why you might want to do this. For instance, if you require your billing to be split by regions or goes, it makes sense for your geo-targeting to mirror this internal requirement. Just note the tradeoff. Introducing geo targeting when you don't need it leads to increased media costs. I'm sounding like a broken record by now, but narrower is more expensive. Broader is cheaper.
I tend to align geo targeting with the natural extent of a brand's distribution. If the brand cannot provide a product or a service in an area, there's little point in advertising there. Geo targeting offers us a way to ensure we don't waste precious dollars.
Targeting only specific devices such as iOS users.
There's a reason you may want to do this. Perhaps your app is only available on the App Store and not on Android devices, you can limit your digital ads to only those users who use iPhones.
I have seldom needed this to target or optimse campaigns. But it's available to advertisers.
Targeting specific placements on websites such as the homepage.
This is the OG of digital targeting. Place an ad on a website in a specific slot for a specific time duration. This slot can be exclusive to you as an advertiser and is most commonly seen on large websites which get a lot of traffic. Here's an example from one of Australia's biggest news sites, the Sydney Morning Herald.
Placement targeting also makes sense on very niche websites like trade websites where you are certain your audience is going to be there.
So if you want to reach search marketing specialists, Search Engine Land will be a good place to run ads.
Placement targeting can be done without the need to overlay other audience targeting options in many instances. If you choose your site carefully you can be fairly certain to reach the right audience.
This is the OG of targeting on digital. It works as long as you choose your site carefully. I would avoid mixing this up with other forms of targeting like audience and behaviours because it can reduce your reach and increases your.... costs.
Why the campaign objective matters a lot
I added this final section even though it is not directly about targeting but I believe it is closely aligned with it. No matter what type of targeting you choose, if you buy advertising using a platform (Google, Facebook etc.), the campaign objective you set matters a lot. With a capital L.
All platforms rely on algorithms that optimise campaigns toward a single campaign goal. Unfortunately, it can only be one goal. Whilst you can create custom conversions which combine different conversion points, the algorithm still optimises towards that one custom conversion.
That means that if you
- Set the objective as reach, it will focus on reach.
- Set the objective as engagement, it will focus on engagement.
- Set the objective as clicks, it will focus on clicks.
- Set the objective as visits, it will focus on visits.
- Set the objective as conversions, it will focus on conversions.
And ignore other things.
This leads to some interesting choices for us when it comes to digital targeting.
We could potentially achieve the same, or better result by simply getting our objective right. Targeting becomes less important.
For example, I have found that targeting a broad demo audience with a campaign objective focused on conversions, can outperform targeting a narrower in-market audience with a campaign objective focused on visits.
The algorithm just has more data to work with a broader audience, and over the life of the campaign, can deliver much better results. Buying to a broad audience tends to be cheaper (see above targeting 101) so your overall cost per acquisition can be much lower simply by going broad with targeting and setting your campaign objective to conversions or leads.
In many situations these days, I recommend my clients to go as broad as possible and set the right campaign objective. Let the algorithm do a lot of the heavy lifting. You may find this works better for you as well.
There are lots of ways to target your digital media.
- Targeting is not the same as digital targeting. Targeting is a part of strategy (Segmentation, Targeting, Positioning) and should be done before you reach the pointy end of digital targeting.
- Get your creative positioning right for that target segment. This is time well spent. 50% of your advertising effectiveness comes down to your creative. Market research anyone?
- With audience targeting, go as broad as you can for that context. It is cheaper. And simpler. Most of the time.
- And finally, geo targeting should align with your brand's limits of distribution.
Ok. That was a doozy. It took me a while to put my thoughts down on a topic which is very close to my heart. I didn't want this to be a curated laundry list so I edited less and kept things a bit more raw.
Have I got all of it right? After 10+ years in an industry that changes all the time, it would be arrogant to make that assumption. I am certain that I will update my position as things change. Every OS needs it from time to time.
But for now, I've found this framework useful to help me navigate the digital targeting space and make good decisions for my clients.
Like what you see? Have a question? Or feel the strong urge to tell me that I'm wrong. I'd love to connect and have a chat about it. DM me on LinkedIn where I am fairly active.