Audience targeting is the #1 success factor for every campaign. While some may argue this point, getting your message in front of the right people is essential for success.
With the abundance of data and tools available today, the options for targeting are endless. But how do you determine what works best for your campaigns and what to avoid? We go from spreadsheets to intensive tech options.
In today’s episode of DGU, Jason and Mark dive into the tools, data, and approaches we’ve used to effectively target the right audience with our campaigns.
Join us as they share our insights and strategies for laser-focused targeting.
Let’s make sure your message is hitting the right people.
If you’re too busy to run through the episode, don’t fret—here are the big takeaways.
PS give the very first episode of Demand Gen U a listen (How To Nail Your Campaign Audience Targeting) if you missed it last year.
Three top takeaways:
Takeaway 1: Bad targeting leads to wasted ad spend
This time last year, the audiences we used in our paid campaigns were much larger.
We tried surrounding entire B2B marketing teams to get them excited at the thought of using Metadata. We could afford to target marketers who wouldn’t use our platform in their day-to-day.
Demand gen marketers are working with tighter budgets in 2023, us included. You can’t go broad with your targeting because you can’t afford to waste anything right now.
You should be targeting three types of roles with your paid campaigns: decision makers, influencers, and end users.
Depending on what you’re selling, decision makers and end users might be the same person too. Now is the time to make sure your targeting isn’t draining your budget.
Takeaway 2: Create audiences that look like your best-fit customers
And when we say “best-fit customers”, we don’t mean your dream accounts.
Your best-fit customers close the fastest, have the highest average deal size, and renew early (and often).
The best way to do this is to create a Salesforce report with closed-won deals from the last six months (new business and renewals). Add in the account field you use to measure the customer health index if you have one.
Compare the profile of these accounts to what you think is your ideal customer profile. Look at the specific industries, market segments, and contact roles for your best-fit customers.
See who attended meetings before the contract was signed. See who actually signed the contract. Sometimes you’ll find things that surprise you too.
Takeaway 3: Use more than demographic and firmographic criteria
Before I started at Metadata, I was lazy about building audiences for my paid campaigns.
I used demographic criteria like job titles and firmographic criteria like # of employees and industries.
I asked my Sales team for an account list if we were trying to target specific accounts. And we all know how well that goes.
Get creative and find unique attributes to layer onto your audiences. See which companies are hiring for specific roles that would benefit from your product/service. See which companies have raised money recently and have huge growth targets to hit.
These are two examples, but you can go nuts here. Using unique data points will help you get laser-focused on the right companies and people.
Which audience do you target?
Mark: Let’s talk about targeting in general because prior to Metadata, I fell into this camp. How did they come up with who they were targeting?
I would try to figure out, usually in a silo, what I thought the firmographic criteria were for the people we should be targeting.
Usually in the shape of industry, company size with employees, location and then layer on some job titles and call it a day. So that was one part.
The other part is, here’s a list of accounts. Sometimes I came up with that list of accounts from a marketing perspective without much research behind it.
But wouldn’t it be nice to have these accounts as customers?
Oftentimes, the more unrealistic accounts were the list that sales gave me of, “Hey, these are Mammoth whales that we’d like to land, wouldn’t it be cool to have some crazy logo on your site and call it a day?”
That was really the extent of my targeting.
What do you think of targeting in general?
Jason: For a lot of us, if we’re in the ad platforms, we stick to whatever they have available and that’s actually improved somewhat over time.
LinkedIn added some different targeting capabilities and then you see Facebook saying “we have business’s targeting now.” Well, no, you don’t.
For most of us, that’s kind of where we cut our teeth on targeting. I think a lot of people assume that ad channels are going to have what you need.
It’s the ad channel and it’s where I’m trying to run ads.
Of course they’re going to have what I need and so you stay in the channel and what their targeting capabilities are within the channel.
You don’t really ever realize you can create your own lists and then upload those into the channel and then add more stuff on top of it.
There’s a lot of other things that you can do. Once you realize that, then the options become open.
Most of us start with the stuff available in the channels, industries, and then we’re left trying to figure it out.
Mark: I remember there were no checks and balances for me doing that. I would just go through a list of industries that sounded like B2B.
Jason: The reality is most of those industries are a mix of B2B and B2C.
If we’re completely honest with ourselves, the ad channels don’t have a reason to help you get more narrowly targeted because then it’s less money that you’re spending.
If I can narrow down the people that have relevance to what I’m saying, that’s less money I have to spend, and then that’s less money getting into the ad channels.
The LinkedIn piece that I hate the most is how they’ve normalized all their job titles now.
People might think that’s helpful. It’s not. It’s detrimental to all of us marketers because now if I put in some niche job title LinkedIn.
Let’s say it looks like it’s a content marketing manager. LinkedIn’s probably gonna take that and bucket it with the marketing manager and then all of a sudden I can’t even target the people I want.
LinkedIn has now made it more broad. Don’t rely on the channel to help you with targeting because they’re gonna benefit from you going way bigger than what you really need. So you have to take it into your own hands.
Eliminating wasted ad spend will improve your targeting
Mark: I think as we were talking about this, there is so much scrutiny on marketing spend right now and that can come in more ways than it probably should.
I think one of the easiest ways to eliminate wasted ad spend is to get better at your targeting.
Walk me through what other VPs of marketing are probably struggling with when it comes to that in the current economy.
Jason: A year ago, the targeting wasn’t as important because a lot of us had three times as much ad spend per month as I do now.
When you have all that money and it’s growing at all costs, you think, let me make my list broader.
What you’re optimizing until that point is to touch every single person that could possibly have a need for my product.
Even if I’m polluting or broadening out, touching a bunch of other people with less relevance because some of these decisions you’re making are targeting, you’re going to leave some people out.
By accident, you’re gonna include some people that you don’t want to include. But it didn’t really matter.
If an audience is mixed, I probably want to cut it out, save the money, and focus on the ones where I like. Especially when I’m using an industry where that industry is like almost all B2B.
Or there is some other qualifier I can put on there, another filter I can put on there to really get to B2B.
There’s really no other way to improve the impact of your campaigns than through who you’re targeting it to.
For example, I’m not in the market for women’s pants. If I were to get ads for that, that would be wasted money on me.
So getting targeted and knowing these are the people that. That’s when you get into— it’s not solely industry because that’s like thousands of accounts.
What are a few ways you’ve built targeting criteria in the past?
Mark: I’ve mentioned what my world used to look like when I first started at Metadata and building targeting lists.
What life was like before Metadata and how I went about it. What are some of the standard ways that you’ve built targeting criteria lists in the past?
Jason: We’ve been talking about industry layering on things like company size, either from number of employees or annual revenue.
Layering on the job title so we’re at the right person. Oftentimes we’ll have maybe several job titles that we want to speak to.
Start with the broad audience. Then here’s one for my CMOs, here’s one for the CFOs, here’s one for the marketing doers, or however you want to split up your audience. Break up into who you want to talk to and how you want to talk to them.
Good questions to ask yourself or ideas to consider:
- Are they on the “INC 500 list” or do they get these kinds of awards?
- Is there a certain employee or revenue band that pops?
- Are they hiring? Are they laying people off? What kind of behavior is going on around it?
Let’s take our existing customers, let’s split them out. Let’s do some analysis on customers and take out the ones that may be churned or that do not have a good health score.
Those are getting a little bit more advanced, but that can help create a list and even something like going into our database and going with everyone who’s at a closed, lost opportunity.
You’re creating segments in your own database by uploading lists into the channels, and you’re starting to target that way too.
Coming up with an ABM list if you want to run an ABM play. That can be done in all sorts of ways, like externalizing a bunch of data and then using that to create the list.
If you’re very sales led, they’ll often have an opinion themselves like, “Hey, no, these are my target accounts.”
I’ve worked at a couple places where our targeting really started from sales.
The sales reps said, “Hey, we’ve done all of our own analysis. Here’s what we’ve come up with.”
Sometimes you take that list and say “ok I’m gonna trust you”, especially if you want leads from these accounts and if they deliver these, you’re going be happy, that might be an easy way to start as well.
All the channels allow you to upload at the very minimum list of accounts and match those. You layer on job titles and things like that. Those are some basic ways.
Mark: When I was at Uptake and we were working in probably five or six different international geos.
I couldn’t tell you anything about the quality of the lists in geos that weren’t the US and even North America in general.
I was just taking all of the regional sales leads word when they said these are the accounts that we should be targeting.
I didn’t know how to target internationally. In hindsight, that was horrible because we wasted a lot of money and I don’t think we really generated much pipeline from any of those lists.
Jason: Geography is often a filter I miss most of the time, multinational companies will have a specific budget. Performance KPIs that are specific to a geography.
You’ll want to split your campaigns by that, usually you start with a big audience type, and then create sub audiences of those.
For example: Here’s my ICP, here it is in North America, here it is in Western Europe, here it is in China, or wherever else.
Most often where that company’s headquarters are located. So that’s what you’re really keying off of, not the leads or employees specific location.
How we created our target account list at Metadata
Mark: Do you remember the big Google sheet that we had when it was a marketing team of two?
The thing that used to crash my computer sometimes because it was so big when I opened it.
Let’s get into how we are going about creating our account list at the end of the day.
Yes, we were using our own product. But also there were ways that we needed to try to get a little bit more creative with finding unique data points to build that list.
Jason: This is where you start externalizing the data and create your own little database and you’re usually at an account level for this.
You’re not getting down to a personal level or a job title level, but at an account level trying to figure out what accounts to target if we’re going to move into an ABM approach or externalize our targeting from these platforms so that we can add more data to it.
A lot of people get hung up on how to merge all these different data sources together.
That can be a nightmare if you’re trying to merge on a company name, because everyone might write a company name differently.
Technically we’re Metadata Inc. But we never write that, I would never write Metadata Inc.
If I was filling out a company thing, so company names are hard to match on, that’s where people sometimes get hung up.
Almost every source of data that I’ve used has a company domain as a field of data, and that is very normalized.
The key value is that the company has a domain name and you’re keying off of that. We had all of these different data sources in our own targeting platform.
Let me start by externalizing that data. I would go in and we have like Bombora for example, or G2 Intent, and I would go into Metamatch and create the broadest audience possible just like Bombora on this topic.
No other qualifiers on it at all. What’s every possible company that has an intent for this topic? Of course Metamatch wants to spit out rows of people like job titles.
I limited it to giving me one per company, because all I really needed is the domain and I externalized that.
I did a little bit of work on it, so I just get, here’s everything from Bombora that they say, all the companies that are in the market for this. Then I did this for a bunch of different data sources.
Here’s what G2 says. If we do a technographic audience on one of the ABM platforms, like 6Sense, here’s what you know. They all spit out a list of domains.
I would take that and start with the basic ZoomInfo information that we’d get from every company.
We’d use some ZoomInfo credits. Then we started to basically do vs lookups of all this data in Google Sheets and like V lookup using that domain.
We would use indicators like yes or no. Because intent is like you’re, it’s on or off. If we found a match, it was a yes.
We did some interesting things too, because like we’d have three different intent data sources. We’d have Bombora intent, slim tell intent, G2 Intent, and so onr.
I’d like to know if they have any of these, but also I want to know if they have a specific intent. Do they have intent in any of these things? That became a new field that we would create.
We normalized things like the number of employees and annual revenue if we had that.
By externalizing the data, you also get to see there’s a bunch of companies that we don’t have revenue data, for example.
In targeting platforms, you don’t see that. You don’t know who you’re missing out on, but now at least you know there are a bunch of these that don’t have it.
Well, maybe I’ll include them too because they’re unknown so I can go with these revenue bands and now that adds to my list. I know I’m leaving one out. I need to remember.
Mark: I asked Jason while we were prepping for this, is there anything that you don’t want to share or feel comfortable sharing with how we built our targeting? And he said, no.
So like you are getting an insight, look at how we do this. It’s not to say that you can copy this, you don’t in the same exact way, because it probably won’t make sense for your business.
But you’ll hopefully get some ideas as to the types of unique data that you can use to build your list.
Jason: I challenge everybody because I’m going to talk about this next in a second. Right now I’m talking about all this data.
I’m talking about sources that a lot of people still know about. Where it gets really interesting is if you can create your own signal data.
We started our first foray into this. Counting ads and getting a count of ads that these companies had on LinkedIn and on Facebook.
You can look at all the ads a company has loaded up on LinkedIn and Facebook. It’s really easy.
You go to LinkedIn, go to their posts from their company page posts, and then once you’re in post, there’s another filter for ads and then LinkedIn. You can’t scroll through and see all the ads.
Mark: It’s much easier now. It’s front and center, but before when we were doing this, it was available.
You had to work a little bit to find it. Now you just go to the company page and it’s right there.
Jason: On Facebook it was even weirder. You go to a Facebook page and you go to like privacy or something.
You follow this interesting other path, but most of them, you end up on a screen that scroll through the ads
At a previous job, I’d worked with this team in Poland to scrape company data off of LinkedIn and create our own company database.
I thought — wouldn’t it be a great target for us? B2B companies that are spending money on LinkedIn and Facebook on these ad platforms?
There’s no data available that I know about that can give you an estimated amount of LinkedIn spend or Facebook spend by company.
These are walled gardens, with memberships. You know people are highly targeted and i think that’s why it’s hard to scrape like a Google search data.
There’s some wonkiness in this ad count because what LinkedIn doesn’t show you is which ads are active in a campaign or not.
There’s a bunch of ads that are loaded up there you’ll count. Ones that won’t be active or old, but I felt like this is a good proxy.
We’d count ads and the system would scroll through, count, and then add it to the database.
We not only use that to determine who would be on the surface that we wanted to laser and get to a target account list.
Not that. But then also the difference between the number of ads they had on LinkedIn and the ones they had on Facebook.
A big part of our play is: Do business advertising on Facebook better than you ever have before?
We could go to them and message people very specifically.
Mark: We had very distinct value props based on, I remember this like based on ad counts in one channel versus the other if they weren’t advertising at all.
I think we grouped it into three different segments with specific value props that we had.
Jason: Sometimes this data you collect, you can use it to personalize your outreach as well.
Or personalize the campaigns you’re running to them. We did a lot of that too. Once we got to that, in the spreadsheet, that’s about where it ended.
It became a beast to manage. And then I had to pay the team in Poland something like $3,000 every time they reran this. That was happening every quarter.
It was like getting out of date quickly. Then a couple times we messed up and corrupted the entire database because we did a V look up wrong.
Here’s actually another piece of advice. You can’t leave the V lookups in place. Once you do the V lookup, then you copy and paste the values because Google Sheets is trying to run that V lookup all the time, for every field. So if you leave them in it never works.
A lot of those are V lookups against a much bigger database that’s sitting in another sheet that has all the ad counts and other info. It gets to be a beast of a sheet.
Mark: Then we were sharing it with our small little sales team and they’re looking in it to figure out how they’re going start reaching out to accounts.
They don’t know how delicate this thing is. People are filtering, they’re hiding, they’re about ready to break it.
I remember whenever their face would pop up in the Google sheet, be like— don’t, don’t fuck this thing up. There is so much behind the scenes that goes into this that you don’t know about.
How our targeting has evolved over time
Mark: Where did you first start once you realized that the sheet was out of control?
I know we’re going to start talking through a couple different data sources and tools that we’ve started to use, but when did that light bulb moment go on?
Jason: Honestly, we took a couple steps back. When we stopped at some point, Britney was using that audience for a while and then it started to become stale. We hadn’t updated it in probably three months.
She started to focus a lot more on retargeting audiences. Then trying to leverage more of the native kind of targeting capabilities in our Metamatch targeting.
We stopped it for a while, but then about 10 months ago, I saw that our buddy Adam Schoenfeld had started a stealth company.
He’s doing account scoring and if anybody knows PeerSignal, they’ve been tracking and collecting data on B2B SaaS companies for several years now.
It’s data that is very rich, it’s updated, and it is comprehensive. I think it’s only B2B SaaS, you have it as one of your high targeting industries. Then it’s a really good set of data.
They actually expand beyond B2B SaaS, but their core model has really unique data and is really around B2B SaaS.
I met with Adam and understood more about what they were doing. You already have a lot of the signals that we’re going and getting manually already.
We kind of see eye-to-eye, but you have a way of bringing all those signals together and then putting them into some algorithm. Then waiting for the different things to actually come out with an account rank and account score.
Being that we were customer number four, we got a pretty smoking deal I think using them. We’ve been using them for about nine months now.
How we use Keyplay at Metadata
Mark: How long have we been using Keyplay?
Jason: Since nine months, we’ve been using Keyplay. Now, Keyplay wants to start almost any Scoring platform.
Analyze your current best customers and your worst customers. Then have a sense of what that looks like and model it off of that.
We knew in our own analysis. There’s no good way for us to tell with any of this data if someone’s going to be a good fit for us or not.
At first we thought, maybe it’s the amount of spend they put through the platform every month. That should be a good indicator. The more they’re spending, the more likely they should be to renew.
Renewals would be a good key indicator for a company to model after. We have these hypotheses, but when we did the analysis, there was literally nothing we could tell in this normal data.
We couldn’t find anything that would easily separate the best from the worst customers.
Then we started entering this recession and we realized even if they’re spending a ton of money now, there’s no saying that they’re going to spend that next month or if they’ll have that same budget. Even a quarter from now.
This is when everybody’s budget started to get tight and pulled back. One of our biggest challenges in our own account scoring is we can’t predict when a company is going to have their marketing budgets cut.
We were having deals in commit, that at the last minute before signing, “my CFO just cut my budget by 50%. I can no longer buy this. I had no zero warning that that was going to happen.”
It became more difficult, so that’s still a challenge we have because most of our platform is tied to ad spend. When that goes away, we tend to go away with it, but we’ve added a lot of data to our modeling with Keyplay and that helps us get around that.
They added the LinkedIn ad counts for us. That’s been more recent, but will be a great improvement to our score.
They’re going to add the Facebook ad count next, but they’ve added things like marketing department size.
That’s not easy to get. Not just that, but then the change to the marketing department size. Often the change in the data is more important than what it is today. That’s either going up or it’s going down.
So now we have a sense which companies are hiring and which companies are making layoffs right now.
We actually will pivot our message because we’ll target both, but we have different messaging for that. We’ll use that data to help kind of pivot that and target that.
Things like the number of job openings or what they are hiring for is what we can start to use in the Keyplay model that helps us. So we’ve been using that and refining it.
That’s where we’re at today with Keyplay. There’s other things that we’re doing.
How we use UserGems at Metadata
Mark: UserGems are something that I know that we’ve been using and I feel like I see them a lot on LinkedIn and people talking about them.
So, UserGems basically try to identify previous champions that you used to work with after they’ve changed jobs and gone on.
So how are we using that outside of the obvious?
Jason: UserGems can spit out a report. So it can just give you a report of all the people that are changing jobs or the companies that have had changes in jobs that are your customers.
We had to wait until that list got big enough for us to create an audience from it. We took it, uploaded it into Metamatch, and created an audience from it.
We’re doing a dynamic audience, so it goes into Salesforce first, and then Metamatch on top of that. Then it updates that audience every day as people kind of come into and out of UserGems.
That’s been a great source because the AEs can use it themselves.
The ads are relevant. They say “ Hey, bring Metadata with you” or something that alludes to the fact that we know they’ve recently had a new job or something like that.
Remind them about the value they got from Metadata when they were a customer.
I think we’ve had UserGems in house for a while, but I think the audiences got to a large enough point where we can use them in advertising. We started that like a month or two ago.
How we use MadKudu at Metadata
Mark: Another one that we’ve been playing around with is MadKudu.
Walk me through what MadKudu is and how we’re using it.
Jason: It’s very similar to Keyplay, but is all the way down to the lead level. It’s not at the account level, but now it’s using all these signals to determine a score.
But it will have some account signals and lead signals too, based on title but also behaviors.
Now you can basically start to leverage, and you end up with a lead getting a grade or a score.
The reps can basically sort by this MadKudu score and see which people at their accounts are actually showing more intent or maybe have a higher score.
We’re looking forward to what MadKudu will do for us is to help us understand product qualified leads, using more of a free trial and product lead behavior to score leads.
People who are using a lot of it and might be ripe for more of an annual commitment or whatever.
That’s really where we want to take MadKudu, it’s a Keyplay for our account scoring, if anything we might just feed the Keyplay account data into MadKudu for the account score.
Then let MadKudu bang on the people’s side of it if we’re not quite there yet.
We have a first model out there, but it doesn’t match Keyplay and we want to ensure there’s a little bit more time there.
Mark: What’s the lift on us to make this all work between Keyplay and MadKudu? It sounds time intensive.
Jason: With Keyplay today, we do the scoring externalize, so there’s not like an integration directly to Salesforce.
We’ll dump our whole Salesforce database in and they’ll score it, then we bring it back in, they’ll score it and add signals to it.
That’s the nice thing about Keyplay, too. It’s not just the score, but any of the signals that we want, like their LinkedIn ad count, we want that actual number.
They can put that back in the spreadsheet and then we upload it. Sam, our Rev op leader, handles all that.
It is a little bit more manual. We’re doing rescoring maybe once a month, something like that with Keyplay.
The platform is pretty interesting, but it gets very scientific with MadKudu. You need to understand some statistics, like nodes, to really be a good user of it.
I’m probably not qualified to use MadKudu. It gives you all this data and it gives you a way to build your models. It’s pretty complicated.
We get help from our CSMs there, but I don’t know if everybody gets that type of help? Do you need a data scientist to run it?
I hope everyone who’s listening learned a lot too. I think the beauty of all this is the first half of the episode we really just talked about how we were using a spreadsheet to come up with who we’re going to target.
Granted, we had access to data sources, but you don’t need a whole lot of crazy tech to do this.
Now we’ve graduated onto using more tech, but everybody’s got to start somewhere and a 29,000 row spreadsheets is where we started.