Before we jump into whether or not to incorporate intent into attribution, let us first discuss where it actually works really well. Third-party data has great potential of becoming incredibly useful and it comes in varying shapes, sizes, and forms. Depending on the nature of third-party datasets, it can positively impact attribution.
When attempting to understand the buyer journey or simply understanding where engagement is and where the opportunities for pipeline are, intent data can be incredibly valuable. It’s always nice to know that an organization or members of an organization are currently interested in a specific solution or evaluating a certain part of their technology stack. So when looking within the scope of a specific tool, when thinking about an organization’s intent, we typically define it under the scope of engagement. I think another term that’s used is search. It’s interesting to note that there is interest out there for a specific organization and whether that is directly a first-party data sort of thing in your website and on your tools or whether it’s searching out there for a solution for that. When grasped at that level, third-party intent can be very useful.
Intent can have this really broad scope and it also can be more narrow. If we use G2 in an analogy where it’s like a garden hose, then Bombora is more like a fire hydrant. However, that is not to say one of them is more useful than the other. It is just the scope of what they can cover is vastly fast and different. And the kinds of data you can get from both is very useful.
We are going to stay still in the land of engagement and understanding the holistic buyer journey. I think both can contribute to a certain chronological event breakdown that you can have particularly if you’re in the B2B space. When you are looking at the organization level, you can see events happening first-party. Therefore, it would be really nice to include events happening third-party.
Typically, these systems give a certain ranking, almost like an indexing of how those organizations are doing for that specific keyword or solution area. That would be great if you have a view to see buying signals there. We’re also seeing really hot, stinking, and intense signals from this organization in the context of G2. You might see an organization looking for attribution solutions, which is a great signal that there might be something that’s getting close to the first-party data even though it’s still happening in G2.
When they’re looking specifically for, let’s say, your organization’s solution in terms of attribution, those should be treated very differently. But that is an example of how you can have very different types of signals. The more I see two world views on attribution and one person may have a share of both, one team often has a blend of both.
One is very marketing-centric, wanting to capture every digital touchpoint to determine what is happening. Then there is the other view who views this as more of a department sliver. How do I describe this as more of a departmental effort? All departmental signals are meaningful and push towards a sale.
The second view I see is this concept of a level of intent gating, whether or not it’s part of attribution and the digital world views. Sometimes, we are just trying to get everything in there.
Unlocking the potential of intent data in attribution: pitfalls and best practices
When it comes to the impact of including something broad as intent in attribution when we’re trying to use it for something like ROI, that is where I get a little bit concerned. First, intent data isn’t always or very rarely that person right by the nature of third-party data. It happens somewhere at the organizational level. It’s all IP matching. “So let’s see if we even know that somebody from an organization is searching for specific terms.” Even if we know that I am not sure that that necessarily constitutes a buying signal. And it’s that part that makes me very hesitant about saying that a would-be campaign member would pass a certain threshold that would say, “Yeah, that’s enough of an activity, enough of a signal that I would want to capture them as a campaign member.”
Maybe, there’s a certain threshold above which you can say, “Okay, we think there’s a strong enough set of signals which we might want to consider as a single event that we can include an attribution model.
Alternatively, let’s say we know that your organization was searched for and there’s a way you can get that data. Well, again it might be at that point, maybe cross a certain threshold which says, “We think it’s a strong enough signal, certainly similar to first-party signals as well. Not all of them should be included. There has to be some thought given to it like, “Well if we were to include it, how can we actually act on it?” “How can we actually take an action knowing it is a certain percentage, can we repeat this in some way?”
Because if you’re going to include high-volume events, then it’s always going to show on top of your attribution modeling, which makes it not always easy to actually take action.
When we are talking about data connectivity and which sources you should consider, it should always start with what people are trying to use the model for when considering which touchpoints to include and how to think about things like waiting. Not waiting for something to happen, but waiting for something more heavily than another.
Intent is absolutely useful, particularly when we’re trying to understand engagement. It may be a critical component of the buyer journey and how we are using attribution, particularly if we’re trying to optimize it for ROI calculations, which should be two different things. Third-party data can be handy for attribution if it’s thought out very critically and only on specific events that pass what I would call a very high threshold of intent to go in as events, knowing that the use of them will be, even at that point, difficult to measure since we don’t know the individual. So it has to be as part of a larger buying committee.
As for engagement, I’ve seen wonderful use cases even with CaliberMind, where we’re seeing an engagement score on first-party data. And right next to it on the event, you’re seeing the intent score, let’s say it’s coming from Bumble. That is a really cool use case because you can have different rates and different indexes from both platforms. And your sales team has that much more information on who they should address when having this view. So certainly for engagement attribution. When it comes to intent, I would be somewhat cautious based on my experience.
As for third-party data, there are other types of third-party data that can be very useful for attribution. For example, data enrichment services built third-party. It’s a little bit out of the scope of intent, but still third-party and very useful potentially when it comes to enriching your data to look at attribution and things like departments and job levels. Therefore, there are different flavors of third-party data and I am not saying all of them should be taken with a grain of salt.
Making sense of third-party data in attribution: identifying valuable signals amidst the noise
There is an incentive to have a different connection to Google Analytics and Tag Manager and be able to catalog those events as if they were first-party. So you’re not potentially losing the data. I would still categorize those as first-party data since they happen in your environment. And typically, with those events, you can identify the person. Also, those events are typically tied to CRM. Yes, it happens on a third-party platform, but that is integrated into your first-party environment. I would still consider them first-party.
When I think of intent, I think we search for specific terms, cookies present across multiple sites and we generally know what the IP is for and so we generally know what’s being searched for. That is a kind of signal where it’s really hard to parse the search for specific areas, to search for the specific area from the actual intent of buying and the actual need of looking for a specific solution. I think that actively on your site or on your social platforms, that’s where it feels much closer to home and I would include those as buying signals and as a part of an attribution model.
At some point, if we look at individual events and say, “Okay, there was a certain visit, and first of all, you’ve got to spot patterns.” so we’re seeing it wasn’t a single impression for a single search, but it’s actually systematic. We’re seeing multiple impressions for the same organization in a short period. We are seeing maybe 10 impressions in a week and all of a sudden, there was probably something driving that. It wasn’t just someone simply searching for a particular keyword. There was probably some sort of conversation. That kind of trend spotting is what I would look at. Additionally, I’d look at whether those are strong enough signals relative to other organizations. Therefore, it’s almost like indexing or normalizing data resale.
We think that this is actually a really high-intent organization. I know we’re seeing strong signals compared to others. So both of those methods and again looking at that in parallel to first-party engagement. I would hope that one would lead the other. Therefore, we might use intent signals then subsequently do some sales motions, and then you start seeing more first-party signals. It would probably be really nice and the hotter organizer accounts would be the ones where you’re seeing spikes in both.
I think that’s a fair leap to make, of course, confirming it with data. I would imagine if somebody’s researching your brand specifically and landing on your site and doing research there as well, that’s a so much stronger intent signal than if someone just clicked on an ad and never actually made it to your site. So ads within those third-party platforms themselves.
I think that both sets of signals could be incredibly useful. I’ve seen the most sophisticated organizations and teams look at both. Then, it takes a process. It takes getting to know the data really well and then realizing we actually get three different types of signals, say from only G2. and they’re not all equal like the ones that are actually happening on our website and we can actually categorize and create these as campaign numbers, even though there’s no automatic communication with your CRM.
We know what the events are and with a flexible enough platform, you might be able to create those as virtual campaign numbers that would speak to an attribution model. Therefore, they could be very, very useful, but will also probably require a good measure of analysis and understanding of what the data looks like in order to incorporate it into attribution. Other considerations include how much data you’re going to be importing into your data warehouse, if you’re building things yourself, and whether or not that’s going to really impact compute.