There’s been lots said and written in recent months about the changes in the world of search. I’ve written about some of them myself (here and here) as it’s a topic that fascinates me. Change is the only constant in our industry and we see a LOT of it on an almost daily basis. If I sit down and think about all the change in the last few years, it could be overwhelming. Here’s just a few of the bigger changes: Alongside these changes, we’ve seen the rise of voice search, predictive search, digital personal assistants. There’s also the growing use of AI and conversations. What does this all result in?
Keywords are dead!
Ok, so maybe keywords aren’t dead. And they’re certainly not unimportant. But they’re definitely not as powerful as they once were. Rather than being the only thing we focused on in determining who or what to target, now they’re simply a facet of our targeting. One of several other factors that we use to hone in on users. The recent changes Google have made to Exact Match serve only to underscore this point. With this in mind, what other sorts of factors should we be looking at? Age, Gender, Device, Location, Interests and Behaviours are all things we’re used to looking at for our other marketing, particularly when it comes to the social side of things, but some of these have never really been key considerations for paid media. Until now. It’s also important to remember that where we are now is only the beginning. I believe things are going to continue to evolve and get ever more complex. In the not-too-distant future I can see it becoming impossible for us to effectively manage our activity without the help of AI and Big Data. There’ll just be too many signals and data points for us to consider when making our bidding decisions and we’ll have no choice but to use some sort of machine learning to optimise. But if that’s the future, what can we do today in terms of using some or all of this data within our campaigns?
If you want to get started on utilising audience data, there’s a number of tools that you’re probably already using within your paid media campaigns. But there’s four in particular I want us to focus on because they’re a) easily accessible to almost everyone and b) you can use these insights interchangeable across different platforms – giving you maximum value for effort. If you wish, once you’ve done some basic analysis you can look at other free or paid-for tools to give you additional insight – but I wouldn’t do that until you’ve proven the value of the basic level of analysis.
Before We Begin
It can be extremely tempting to start analysing the data we’re going to look at down to the nth degree – it’s like disappearing down an endless rabbit hole and if you’re like me it’s easy to get lost in the detail. I’d encourage you to think of the 80/20 rule here (cliched but it works) – focus on the biggest, most actionable insights and work on these. The data you get needs to actually be usable within your campaigns and if you’ve drilled down to an audience that contains five people, that’s not usable. Bear in mind that to use these insights across Google, Facebook etc, you need minimum audience sizes – so perhaps don’t drill down to any audiences smaller than a thousand users. Let’s get started!
Google Analytics is great because it’s free (unless you pay for Premium) and it gives you all of the data signals that we’re interested in: Age, Gender, Device, Location, Interests and Behaviours. Because you can also overlay this with your conversion and on-site data, you can get really powerful insights into things like:
- Converters vs Non-Converters
- Repeat Customers
- Most Valuable
With things like the above you’ll really be able to drill down and get really detailed data. The first place to look for your basic insights is within the Audience section of GA. In here you have reports for all of the audience elements. It’s worth paying particular attention to the Interests report as it gives you two specific lenses to view Interests through; Affinity Audiences and In-Market Segments. What are they? Affinity Audiences let you identify someone in terms of their lifestyle – what content are they regularly browsing online? What are their hobbies likely to be? In-Market segments are a little different – they identify users in terms of their product-purchase interests/intent – for example are they in the market for a new car right now, or travel insurance? What can we infer this audience wants to buy in this moment? Depending on where in the funnel you want to target users, both have definite value.
You can get quite in-depth with the data from GA. As well as using Secondary Dimensions to layer age and gender, I’d also suggest using segments – not just the ones Google creates for you – but custom segments. Take things like age and gender and create them as segments for your biggest user groups, like in the example below. Then take these segments and combine them with the other Audience reports in GA – overlaying information such as:
- Channel used (PPC, Social etc)
- New vs returning
- Converters vs non-converters
This will begin to give you a rounded customer profile – does the Facebook visitor look different to the Twitter visitor? Are your PPC converters different to those who come in via email? There may be no difference, but it’s good to examine. Start to layer on the insights to develop a wider customer profile. Try writing customer snapshots!
“Women who convert are more likely to be interested in movies if they’re between 18-24, but if they’re between 25-34, they’ll be news & celebrity junkies. If they’re coming from email or Bing, they’re likely to convert better.”
Try writing snapshots for some of your top user segments.
If you run an email programme in your business you have a ready-made source of great data. Find out what data is available in your platform and what segments might already be in place. Then if you can, get downloads of the email addresses. Make sure they’re somewhere super secure so that you comply with EU data protection legislation on the use and storage of customer data. If you’d like additional ideas on segments to look at, I have some I’d recommend – in addition to profiling the whole database:
- Converters & Non-Converters
- Top/Most Valuable Customers
- Repeat Customers
- Sale/Promotion Driven Customers (do you have users who only buy when you’re offering a discount?)
- Break out by product type
- Most Engaged Users
- Organic vs Promotional Signups (if you acquire email addresses via third parties or promotional means such as competitions, does performance/audience make-up vary compared to those who sign-up organically?)
Once you’ve downloaded the addresses and stored them somewhere safe, you’re done – all the hard work happens in the next step.
While you can get audience insights from Twitter, we’re going to focus on Facebook as the data found here is richer and more accurate. We also have the bonus of being able to upload our email data to understand the makeup of those user lists. Top Tip: You’ll get the most value from Facebook if you have the pixel installed to track users and conversions on your website. If you use GTM it’s relatively straightforward to set up and doesn’t slow down the site. You’ll be able to combine audience and pixel data for conversion points such as purchases, registrations, searches, add-to-cart etc, for a level of insight very similar to what you can get in Google Analytics. There’s two key areas to focus on in Facebook Ads Manager (make sure you have access to this to obtain your insights) – Plan > Audience Insights and Assets > Audiences. We want to upload our email data first of all, so we’ll head to Assets > Audiences. We have 3 options when we create an audience:
- Custom Audience – upload your emails here! This is also where you can create audiences from your pixel data if you want to select site users, or people who visited certain categories/completed certain action. If you have a Facebook Page, or have previously run Facebook campaigns, you can also use Custom Audiences to select users who’ve interacted with your Page or past campaigns by watching a video, filling in a Lead Ad, etc
- Lookalike Audience – you can choose a Custom Audience, Pixel or Page and find people like those people. You can also specify how closely you want them to match your existing audience which will influence the number of users you get in that audience – the closer a match you want, the fewer the number of users in the group and vice versa
- Saved Audience – this lets us use Facebook’s targeting to create a group of people and is also where you’d go to combine different Custom Audiences or Lookalikes together. This would allow you to find users who’ve interacted with your Page who also converted. Or non-converting users on one of your specific email lists, and so on
We then move to Plan > Audience Insights: Given our work on creating Custom Audiences just now, we’re mostly interested in the third option. It’s not that people who are connected to your Page aren’t important, but you could create an Audience of people connected to your Page who also took the desired conversion actions on your website – since initially, these might be the people you’re most interested in, rather than just all fans. Once you choose an option you get this fabulous screen filled with data! You’ve got multiple tabs across the top and loads more information below the fold that you can explore for each of your audiences. Not only do you get your absolute numbers, you can also see how they compare to the average on Facebook – a great way to understand whether you’re potentially missing an opportunity with certain key interests or audiences. All you have to do is select one of the audiences you’ve already created, or enter your Page and it populates for you. You can get so much information! Some of the more advanced insights aren’t available if you’re not looking at audience data for the US, but it’s still immensely powerful. Given the reach of Facebook’s platform and how much data people have volunteered about themselves, in some ways it’s a better data source than GA or AdWords which rely on some assumptions. Remember you can get insights on:
- Page Likes
- Relationship Status
- Job Title
- Visit Frequency
As with GA, I’d encourage you to take your top-level insights, save them as an Audience and then use this saved audience to dig deeper – do the page likes or visit frequency to Facebook differ between your 20-somethings and your 50-somethings? How about their device usage? Remember to layer the insights, but not so much that they cease to be useful.
Finally, onto Search Insights from AdWords. Within these Insights you can get data for both your campaigns and any pre-determined audiences. Let’s head to the Shared Library > Audiences: Any Audience that has been historically created in AdWords or GA (if you have them linked) will show up in this list. AdWords has some lists that it creates by default and if you have no other lists at this point, they’re a good place to begin to look for audience data. Click on any of your lists that appear here. Straight away you’re presented with Top Insights and a chart showing you the growth of your list size from the last 30 days up to 6 months ago, if your list is that old. Along the top, you can start looking at the detail, including In-Market Audiences and Affinity Audiences, so you’ll easily be able to see makeup of your audience based on your Converters list, or the All Visitors 30 Days list. This will support the data you could get from GA, but is a great source of information if your business doesn’t use GA. What this does give you that GA doesn’t is how your numbers compare to the average of the UK (or whichever region you specify) and the amounts of impressions available on a monthly basis if you targeted these users via the GDN. You can also get other insights such as Device, Age, Gender and Parental Status. You may notice with some of these numbers particularly around age/gender/parental status that sometime the data comes back with “Unknown” rather than a value. This ties in to the assumptions that Google has to make about its users. These assumptions are made from specific site data. Sometimes site don’t/can’t/won’t share that information with Google, leading to a gap in the insights. This is where Facebook can have the edge as generally users volunteer this sort of information freely. If you’d like data at a campaign level, you can get that too. If you navigate to “All Campaigns” > Audiences you can start to gather the data if you select the “Demographics” label: You can select individual, multiple or all campaigns to look at. You can also choose to see the data sliced by Clicks, Impressions or Conversions in the drop-down that appears above the graph – it can be useful to look at all 3 views. This data isn’t as full featured as the data you get for audiences as you’re missing interests and parental status, but given you can only apply insights of this kind to your search campaigns at this point in time, it’d be a bit of a tease of Google to give you any more. If you want, you can see what the demographic combinations look like for individual ad groups by selecting this slider which appears below the charts in the bottom right-hand corner. Seeing the data broken out like this should give you what you need for applying any bid modifiers, should you want to bid differently by age and/or gender. That’s it for gathering our audience insights, which is by far the hardest bit. When it comes to applying them within our paid media campaigns, that part is dead easy!
Using Your Insights
Before we begin, here’s a reference on what you can apply where, based on the data you’ll have from your research phase. Which channel should you use? It depends on your goals! Do you want to drive sales or leads? Increase brand awareness? Advocacy and retention? Let that guide you. Remember you don’t have to use individual channels in isolation – you can use them in tandem, depending on your goals and I’d suggest using more than one, if not all of them for a more full-funnel-led approach.
Age & Gender
You’ll now know the performance of age and gender by segment. You’ll be able to alter your bids at ad group level, having positive or negative bid modifiers relative to performance, or you can exclude/remove elements from your campaigns entirely if they don’t work for you or aren’t relevant. A few notes on applying your changes:
- “Unknowns” – Using the charts we looked at earlier, you’ll know how big a chunk of your potential audience is “Unknown” – Google’s usual recommendation is not to exclude Unknowns, if you only want to target a certain age or gender, particularly if it makes up a large chunk of your audience. Something to bear in mind!
- Parental Status – you can’t utilise this data currently within your search campaigns. It’s only available to use for Display and Video
- Household Income – this is currently a US only feature and allows you to target by income bands
It’s important to note that these can only be utilised in Display and Video campaigns, by ad group. They appear under the Display Network or Video Targeting tabs. The Interests data available from GA, where both Affinity Audiences and In-Market Audiences are available to analyse, or our Adwords Audience Insights can be used here. You’ll notice there’s also an option for Custom Affinity Audiences – which I’ll cover off in my next article where I’ll talk about some expert/next level hacks to help you go even further with your audience analysis. Topics is a little different – this is literally the content topics of the pages where the ad units are. It’s not specific to a given audience, intent or behaviour, but it would be great if you wanted reach and impressions. It can work alongside some more focused data or ad groups within a bigger campaign, particularly if you’re trying to drive brand awareness.
Once you know who your best converting audiences are, or alternatively your worst converting, you can create dedicated audiences in GA that will be available for you to use in AdWords. You can also create lists or custom combinations of lists within Adwords itself. A Custom Combination would be something like All Visitors, excluding Converters – to target only people who are coming back to your site who haven’t converted. Once you’ve got your lists, you can apply or exclude them across both Campaign and Ad Group level. This gives you the option to stop bidding on keywords for a specific audience that you feel doesn’t work well for you or alternatively to bid more, should a customer matching your desired profile search for one of your keywords. Google’s new “Similar Audiences for Search” feature can help you with this too, if you wish. When you apply an audience you have 2 options:
- Bid Only – this allows you to simply bid differently should somebody on one of your lists shows up
- Target & Bid – this allows you to only bid if the user is on one of your lists – great if you want to show a different message, or allocate a different budget to new vs returning users
The recent ability to apply/exclude lists at Campaign level is HUGE – this makes managing your account much easier and helps reduce the number of elements, as long as you don’t want to add/exclude audiences within the Ad Groups in those campaigns. Currently you can only have Campaign OR Ad Group audiences – you can’t exclude at Campaign level and apply at Ad Group level.
Let’s take a quick look at Bing!
Age & Gender
Bing has offered Age and Gender bid modifiers for ages – an area they’ve long set the lead in. You can get Bing-specific data from GA by applying a segment that covers your Bing CPC users when looking at the Age & Gender reports in Audience. You can then turn these findings into bid modifiers and apply them. You can’t exclude a grouping completely, but you can apply a -90% bid modifier, which should exclude most unwanted users. You can also bid up to +900% for a positive demographic.
Unlike Adwords, Bing Ads doesn’t have a sync with GA. What it does have though, is the Universal Event Tracking (UET) tag. You can use this tag like Facebook’s Pixel to set up and track multiple conversion goals as well as tracking All Users or visitors to a specific part of the site. Once the UET tag is enabled, you can create custom audiences and assign them at Ad Group level. As with Adwords you have the options of “Bid” or “Target and Bid” and you can see +900 to -90% bid adjustments. Exclusions are also available, should you want to exclude a specific audience entirely.
While Facebook is perhaps the platform with the most data to get from it, once you have that information it’s remarkably easy to apply. You could either select one of your Custom Audiences from earlier, try a Lookalike Audience, or build a targeting profile from scratch based on the insights you’ve gathered. Chances are you created some Custom Audiences, Lookalikes or Saved Audiences during the research phase but if you didn’t, now is the perfect time to try them out. That’s it!
It’s easier than you might think to start using audience data within your paid media campaigns and you can start using the tools and platforms you use every day. Top tips to bear in mind would be:
- Don’t get lost in the data – focus on your biggest, most valuable audiences and your key insights
- Make sure you’ve got the Facebook Pixel installed to get the most value from that platform
- Try and humanise your customers by coming up with little user snapshots to help give them a personality
I’ll build on the above with my next article which will feature some expert tips and tricks to take you beyond the basics and to the next level when working with your audiences. Do you have any hacks you’d like to share? Let me know in the comments!