The proper way to approach FB ads targeting.

I do a lot of Facebook ads ‘audits’ and one of the biggest issues people have with Facebook ads is the targeting.

There are a million options, it’s all confusing as hell, and you don't know where to even start.

Breathe.

The targeting available through Facebook ads is possibly the thing that holds it above any other advertising channel as the most effective place to put an ad out these days.

So here is a brief-is explanation of how to go about targeting your ads.

I’m not including every single detail because I think it would get too confusing, and that’s the opposite of the point of this.

First up, here are the available targeting options, and what they all are:

1. Custom audiences -

These are audiences built from user behaviour. 

• Your email list (or a section of it)
• People who visited your website, or a specific page on it, or took a specific action (i.e. filled in the contact form)
• People who have interacted with your app.
• People who have interacted with something on your FB page (a video, a post, a lead form, an event) or your Instagram page.
• Or you can set offline interactions, but I’m leaving that out here for simplicity’s sake.

The pixel that you have associated with your Ads account tracks all this stuff, and you can tell FB that you want to create any audience listed above. If it’s something like ‘people who engaged with this video in the last 7 days’ then that will be a constantly updating group of people over time.

We use custom audiences for retargeting, but that's not something I'm covering here.

2. Lookalike audiences (LLAs) -

This is where you take any of those custom audiences you created, and tell Facebook “find me more people like this”.You choose a country to base your LLA in, and a percentage of the population - so if you choose 5% and Belgium, then Facebook will make a big list of the 5% of Belgian people it thinks are most like the people who watched that video in the last 7 days.

3. Location - 

List as many locations as you like to make up your audience. Also add areas to exclude if you know those people aren’t who you want to see your ad, and you can specify people who are:

• currently in a location,
• who have recently been there,
• who live there,
• who are travelling there,
• or all of the above.

You can also drop a pin and choose a radius around that point it you want.

4. Age & Gender -

Fairly self-explanatory, although there will always be people in your results who haven’t told FB that they are one gender or the other.

5. Interest targeting -

The area that people get most wrong, most often - and the subject of a lot of focus with all the Cambridge Analytica stuff.You can choose to target people based on 

• Demographic info - such as ‘newlyweds' or ‘college educated'
• Interests - such as ‘handbags’ or ‘volunteering'
• Behaviours - such as ‘owns amazon fire’ or ‘listed themselves as small business owner'

You can narrow your audience by choosing people who are in both interest A and interest B, or by excluding a group - so people who like A but NOT B.

6. Connections -

Choose to target or exclude people who already like your page/event/app, and/or their FB friends.

- - -

All that choice. Where do you start?

You start with the end in mind.

Facebook is a very powerful platform, and you always want to be setting things up in such a way that it lets the mysterious algorithm do the heavy listing.

For that reason, you’ll find that 9 times out of 10, lookalike audiences work the best.

So, work backwards from there. The first question you want to answer is - Do you have a good enough custom audience to build a Lookalike?

You can’t create a lookalike audience from a custom audience of less than 100 people, but also you wouldn’t want to.

Facebook doesn’t give us any official numbers either, but from my experience, you want at least 500 ‘buyers’, 1500+ email subscribers, or 5000 'page likers' to get the best results.

This is always a judgment call - it makes sense that 500 current clients is better than 1000 people who just watched a video on your page - but where exactly those cut-off points are we don’t know.

- - -

If you don’t have a big enough custom audience, then you’re likely going to look to the interests and age/gender/location to get you there.

Tips for non-lookalike-based targeting:

1. Don't go too broad, if you run a yoga shop, then adding ‘health’ and ‘hobbies’ and ‘sport’ is going to include literally millions of people who aren’t a good fit for you.
2. Make use of the ‘exclude’ option - so if you wrote a book called "how to sell gently", then adding in something like ‘entrepreneurship’ and EXCLUDING people who like Grant Cardone could be worth trying.
3. But don’t go too small - unless your budget is tiny, targeting business owners who have expressed an interest in both Kundalini yoga & meditation, is not going to leave you enough people.
4. Don't lump together a bunch of interests in one ad set, because you can’t then tell which one worked and which didn’t - split them into different ad sets
5. Testing is key - you might find that targeting people who like Tony Robbins works great for your online meditation training, but there is probably a cheaper option to target out there.
6. If you are a local business, then forget all about interests, and just stick a radius around your business location that you know people already travel to get to you, and add in the correct age-range and gender.
7. Get very clear on your customer avatar - the better you can describe them, the easier it’s going to be finding appropriate interests to test.
8. Make use of the ‘friends of people who like your page’ option - social proof is a massive deal when it comes to persuasiveness, and FB tells you that “Steve like this page” when they show you the ad, so use that.

But the big point is to TEST.

• Don’t be the person who threw an ad up, left it at ￡10 a day for a week, then said that FB ads don’t work - you’re a fool.As the platform gets more saturated with ads, the costs are rising, and often winning or not with FB ads comes down to being able to find pockets of audience that aren’t being reached by your competitors.
• FB ads are an auction, and the amount of people who want to target the same user is one of the things that determines the cost of your ads, that’s why going for ‘people who like football’ is expensive.
• Use the ‘breakdown’ column in Ads manager to look at the different delivery categories and try to find groups of people you can try excluding.
• A quick point on analysis - telling people to test is all well and good, but HOW do you do that.
• The easiest thing to look at is the CTR (click-through rate). 
• If you’re testing two audiences against each other, with an idenctal advert - then the amount of people who click on that advert is a safe metric to use to know which is a better fit.
• BUT - don’t focus on that at the expense of conversions - if you’re trying to get signups for your ebook - then look at the cost per signup alongside the click-through-rate. It’s find to pay 87p per click compared to 52p IF those 87p’ers are converting twice as well.

- - -

Here are some Extracts from audits I’ve done for various businesses in the last few months to give some real life examples:

Charity organisation offering low-interest loans - 

"As you can see, the generally high relevance score and CTR shows that the targeting is relatively strong, but there are ways to improve it I'll cover later in the report.

Furthermore, there is no significant return (ROI) in targeting 35-44 year olds, so they could be excluded”

Local football coaching program for kids -

"It’s hard to be definite on all of this because the way these ads are setup means I can’t see number of signups, but it looks like your targeting could probably be improved.

Location: You are using location targeting which is great, I often get better results using a radius around an area that specific towns, but that is just something to test, rather than a problem.

Age range: You are showing ads to <16’s all the way to 60+ year olds. This is likely costing you money. Have a look at your existing clients, I bet that 80% of them fall into a 20 year age range, maybe 30-50? If so, your money is going to be best spent just speaking to those people.

Gender: Without testing I’m couldn’t say for sure, but might these ads to work better to just women? Typically they are the ones organising activities, and even if (to generalise massively) dads would be excited about their kids going to football classes, it will still probably be the mums who sort it out."

Online coaching company - promoting B2B events in London -

"For any type of local event, I’ve generally had best results from fairly open interest targeting, but quite tight geographical and age targeting, so I would try that, rather than something like London + 62km, which is a massive distance/amount of people.”

Driver instructor training -

"For your traffic campaign, you’ve left the targeting blank, other than using age and location.

This is often an effective strategy if you serve a wide client base, but you’re using a 5km radium, which seems very small, and again means that the frequency is very high (6/7+ on those ads).

I don’t know enough about your business to know if 5km really is the furthest people will travel, but it seems very tight.

In terms of the age groups, you are getting the best click through rate from 25-34 year olds (males specifically), which suggests the messaging is connecting with them best.

In general, men are clicking more often than women - which doesn’t seem surprising”

SEO company

"You’ve created a saved audience that comprises of people who are identified themselves as small business owners, and are interested in SEO stuff.

That’s a perfectly good starting point, but leaving it there means that you’re starting from scratch each time.

Using lookalike audiences, based on people who have read/interacted/ signed up with you before will let FB’s algorithm do much of the hard work and should see better results.

If you also follow the above strategy, you will be creating audiences based on past behaviour and interaction, which means you’ll be showing your stuff to people who have already shown an interest, rather than showing it to brand new people each time and letting this interested people go cold.

How is the split in terms of age/gender with your clients?

Looking at your ads, you can see that there is often a big different in cost per result for male vs female, so this is worth paying attention to, so that you can get the best ROI on each campaign.”

Low-price gym chain - 

"The temptation is to add in interests like ‘Weight loss’ or ‘Crossfit’, but honestly, it’s not the right way to go.

It’s more expensive to target people who have identified themselves as interested in specific things like Crossfit because FB works like an auction, so choosing an ‘obvious’ interest means that everyone is bidding on the same people, and the price goes up.

Also, you’re missing out on all the people who don’t already ‘like’ fitness related topics yet, but would be open to improving their health & fitness.

- You’ll likely get better results just sticking to age, gender and location. Have a look at where your clients come from, how far away it’s worth travelling, and choose the radius around your base according to that.

- Your messaging will then qualify people, rather than using the targeting options."

Ecommerce business selling gaming console products

"For some of your campaigns, you’re doing no targeting at all apart from UK, aged 25-44 - that’s an audience of 20m people, and clearly your ads aren’t relevant to all of those, so you’re getting charged more because of the amount of people who don’t care.

The thing is, computer games are some people’s literal FAVOURITE THINGS, so when you get your targeting right, all the metrics that Facebook care about (relevance and engagement mainly) will go through the roof, which will see your ad cost plummet.

Basically, you’re not testing different targeting options, and you’re not using the data to guide you.

Age: You want to be testing different age ranges for different campaigns - firstly think of who each campaign will be most relevant to, but then look at the numbers - this [XYZ] ad for example: You’ve spent ￡100 to get 2 sales from 25-34 year olds, and ￡157 to get 17 sales from 35-44 year olds.

Gender: Clearly your audience is mainly guys, and the numbers show that, with conversions and click-through-rates way lower for females, so stop spending money on them.

Interest targeting*:* With the level of fandom that people have in the computer game world, you’re leaving a big opportunity on the table by not testing interest targeting.

That said, given that you’ve now got a sizeable number of clients and FB page fans, you might be able to jump straight into Lookalike audiences.

Lookalike audiences: Using your existing clients & fans as a seed audience you can basically let Facebook get on with what it’s best at, which is letting the algorithm run.

The way to do this is to start broad and get progressively narrower.

Friends of FB page fans*.* I saw you tested this, which is a good idea, but think about using that information to get engagement, i.e. who did you used to play [x] with?or take comments like this and run a “does your boyfriend love old computer games...” type ad:?"

Online health coach - 

"Telling Facebook you want people who match ANY of the following targeting options: Food & Drink, Entertainment, Hobbies & Activities, Technology, is far too broad, it’s basically everyone, who DOESN’T like entertainment or hobbies…?

Instead, narrow the audience down, so people who Like X AND Y AND Z. That starts painting a picture.

But actually, I think you’ll get more benefit with interest targeting if you think about the BUSINESSES/ PEOPLE who your target audience follow, and combing them for your targeting.

You sent me the excel file with all that good avatar info, so try out women who like 50 shades AND Secret Escapes & Come Dine With Me...

That will get you started, then once you’ve built up enough people you can play with the Lookalike audience feature that FB has."

- - -

This ended up longer than planned, I hope you find it useful - don’t let the complexity of FB’s targeting options put you off.

Post up any questions in the comments.