An Amazing Best-Practice Guide with Original Examples of Lead Scoring in Autopilot
Bold promise. But it works.
Lead scoring by itself is revolutionary. You’re taking something that once required extensive human attention and automating it, which nearly always adds massive value. And you’re building a set of rules instead of relying on well-I-reckon instinct and feel. Again, the results from that are almost always massively better.
But lead scoring can do a lot more than it’s credited with. It can give you automated, granular control over your whole lifecycle and power the process of providing your customers with an experience from your company that feels tailor-made for them (because it is). And it can work for any company — big, small, B2B, B2C, SaaS, or engine parts or ecommerce clothing store.
How do we know? We wrote the book on it.
Before we get into all the cool sh*t you can use lead scoring for, let’s very briefly talk about what it is. If you already know, skip this bit. If not, come this way.
Lead Scoring: What Is It?
Lead scoring is the process of assigning a numerical or letter score to a lead. The score is based on things like behavior and identity attributes. When you add up all the numbers for who the lead is, where they work, what they’ve looked at on your site, and a bunch of other stuff, you get a score for that lead. If the score reaches a certain tipping point, you automatically take action.
Most lead scoring systems use 100 as their tipping point, and most use “tell sales to call this person” as their action.
As I said, just by itself, this takes a lot of fluff out of the process and replaces it with a replicable, data-driven, if-this-then-that process. Totally in favor of that. A typical lead scoring model might look something like this:
|2+ years in business||+25|
|SaaS B2B business with at least 1 high-value client||+50|
Uses a rival service
|Recently stopped using a rival service||+50|
|Has $1m+ ARR||+25|
|Uses a productivity or communications suite your MarTech tool stack integrates with||+50|
|Uses a productivity or communications suite your MarTech tool stack doesn’t integrate with||-50|
|Referred by an existing customer||
So, if person A recently stopped using a rival service and was referred by an existing customer, they get a call today because they score 100. So does person B, who has $1m+ ARR, is a SaaS business with at least 1 high-value client, and has been in their business 2 years. They score 100 too.
Person C was referred by an existing customer (+50), uses a productivity tool that your product doesn’t integrate with (-50), has $1m+ARR (+25), is a B2B SaaS business with high-value clients (+50), and uses a rival service (-50). They score 25, so they get put on an email list.
Before, it might have taken an experienced sales person looking through these leads and weighing them to arrive at this conclusion. Get lead scoring automation set up and it just happens. Sales isn’t trying to sip from a firehose, marketing isn’t warming up hot leads, and you don’t lose your whales in a fog; when the kinds of leads that deserve to have all hands on deck heave into view, they get flagged early, automatically, and accurately.
Lead Scoring Can Do a Lot More than What You Find in 101-Level How-To Guides
As valuable as the above strategy is to everyone in lead generation, it’s the tip of the iceberg of what lead scoring can do.
Lead scoring works because it assigns numerical values and then automates actions based on those values. It bears repeating because out of its context you can see how you could use it for other things.
If you have any decent marketing automation software, you should be able to build everything-scoring automation.
How does everything-scoring automation differ from ordinary marketing automation? The difference is in the numerical system that lets you feed variables into the system, calculate a score, and use the score to execute automated marketing action.
Lead scoring automation (or our idea of everything-scoring automation) is a lot subtler than an if-this-then-that marketing automation system. The latter only uses one behavioral signal, such as cart abandonment, when it moves people over to a specific cart abandonment email list. Lead scoring automation can take into account that person’s previous behavior and their identity when it decides which messaging within your marketing strategy to send them; and it can do it automatically, and without using some super-specialized AI software that doesn’t really work.
Here are some of the things you can use this everything-scoring automation for:
You Can Use Lead Scoring Models with Both B2B and B2C Generation
Stereotypically, lead scoring is a B2B tech company thing. B2C companies don’t really get a lot of attention in most of the blog posts and ebooks and videos that talk up lead scoring 101. But they should. Lead scoring, and the everything-scoring that you can build on the same framework can have transformative effects for B2C companies.
Whether it’s B2B or B2C and whatever the offering: if you can measure it, you can score it, and use that score to automate the right action for the right customer at the right time.
For instance, suppose I’m running a B2C company and someone comes to the website and leaves their email — along with a few thousand others. The difference being, this particular visitor is adding products to their cart, visiting multiple pages in different areas of the site, and generally making themselves at home. They’re giving off all these signals that they’re way, way more interested than the average visitor. That is, they’re not just more interested in making a purchase; they’re more interested in the business, and we know that because they’re spending a lot of time on different pages on the site and showing us in all these other ways (that we’ve spent years honing the ability to detect).
This person needs to get different treatment than Jane Doe, Every visitor. These are the kinds of visitors that Levis would send info about vintage selvage to, rather than just hope to sell a pair of 501s: they’re really into it. Which makes them vitally important to the business.
This image illustrates something that most successful eCommerce stores find out from their own data: a fairly small subset of customers is actually generating vastly disproportionate revenue, and they’re also far more likely to be serial customers.
In this case, 5% of customers generate 17% of total revenue, with each of those customers being worth around 3.4X the average, and the bulk of that coming from far-larger-than-average orders. A sale to one of these people is worth far, far more than the average, so it justifies far greater efforts, but first, we have to identify these potential customers as early and accurately as we can.
So we have several methods that we can use — segmentation, personalization — but they tend to rely on yes/no triggers. People who put stuff in a cart but didn’t buy it, they get cart abandonment emails. When this is done really well, they get emails relevant to the thing they put into their cart but didn’t buy.
Which works a lot better than just hoping they come back. But it’s still a pretty blunt tool compared to using lead scoring to establish a numerical picture of the potential value of this customer.
What we’d do is assign those behaviors a numerical value, so we can get a much clearer picture of the kind of customer we’re dealing with. Someone who’s put seven items in their cart, viewed twelve web pages, and spent 30 minutes on site gets assigned numbers for all those activities, with higher numbers for more pages, more items, and more time.
Suddenly we’re beyond if-this-then-that for individual behaviors, without falling back on identifying these customers by human judgment. We can send this high-value customer to our high-effort email and social contact lists, effectively giving them the penthouse suite of our marketing efforts, and we can do it accurately, instantly, and automatically. Just because the action you take isn’t “get sales to call” doesn’t mean it’s suddenly not lead scoring.
You Can Use Lead Scoring for Customer Success
We’ve talked about positive lead scoring: adding a person’s score up until it gets high enough to trigger an action. But you can invert that model and use it to power your efforts against churn. Then, using the same score-everything model you’d use for positives, you can broaden the scope of that approach and use it for customer success.
Here’s how that works with the framework outlined above:
|…not used the application for a month||-25|
|…never used one of the core features of the application||-50|
|…visited the unsubscribe page||-100|
If they get to -100, they’re targeted for special assistance meant to stop them from churning.
The first thing you’ll notice about these metrics is that they’re available from the engagement data that any SaaS company generates and watches as a matter of course. And the second thing will be how simplistic they are. These types of actions are deafening klaxons. If automating responses to them was the best that we could do by bringing our score-everything approach to customer success, it would still be a step in the right direction. But by itself, I wouldn’t call it a game-changer.
What does make it a game-changer is the capacity to use that scoring process to make it far more subtle? We don’t have to have -50s and -25s. We can use engagement metrics that are way below the actionable threshold to build a more accurate, nuanced picture that can still be automated.
You Can Apply the Framework in SaaS and Reduce Churn
Faced with unsustainable churn, Groove went back over their data to see what they could do to improve customer success and found some obvious things — and some more subtle tells. Groove refers to these as “Red Flag Metrics,” and they included time spent creating new rules and customizing support widgets. When they found that customers who churned spent far longer doing those things than customers who stayed, they used time spent as a Red Flag Metric. We used similar methods years back when we set up lead scoring for Kissmetrics, with an assist from Mixpanel Predict.
The big difference between the two approaches is that we’d feed that first-level information into a scoring system that was weighted to indicate how at risk the customer was, and automate the right response based off of that. Do this right and you can take the engagement scoring system that a SaaS company is already using, and build lead scoring so it syncs perfectly along with it and uses it as the source of its most important data.
You Can Use Lead Scoring Models for Ecommerce
Ecommerce sites can use the same methods that we’ve already talked about. You just put different numbers into the equation: instead of tracking app engagement, you’re going to track content consumption, page views, and purchase behaviors. Suppose you get a user who visits the site, adds six products to the cart, then buys them all. Great, right? And if instead they leave and abandon that cart, well, you probably have automated cart abandonment emails set up, so you’re already working to get back that customer. This kind of either-or stuff is already pretty well covered.
What about if they put six items in the cart and buy five? That’s the kind of neither-nor territory that automating for binary outcomes doesn’t even touch. But how many times have you removed just one of several items from the cart yourself?
We can track a user through an ecommerce site from their first visit and even offsite using UTM tracking, and assign numerical scores based on content interactions, on repeat visits, on email opens — on everything someone can do or not do relating to an ecommerce business.
If that user then puts those six items in the cart and buys five, we’re ready to follow up with specific, personalized messaging. You’re moving people over onto custom audiences and more active email lists, tailoring your retargeting and abandonment efforts to where you know the money is going to work harder. You can offer targeted discounts. Basing all this off of your data should mean you spend more on some leads, but overall CPA goes down. This isn’t what people talk about when they talk about lead scoring, but it should be, because this absolutely is lead scoring.
Download our RealThread case study to get full detail about the lead scoring process of a B2B ecommerce client.
How We Do It – Lead Scoring Automation in Autopilot
We started being able to do this when we set up Autopilot, now our technology partner and content partner. The standout feature of Autopilot for us is a one-to-one integration with Salesforce, which meant we were able to create journeys based on custom engagement and lead scores that were fed from Salesforce and were informed by them.
From lead scoring, we moved seamlessly onto scoring across the whole customer lifecycle, so anyone we’re in contact with is always being automatically put in front of the right eyeballs. This example of lead scoring application is only possible if you track data, otherwise you have nothing to work with. The idea that a bunch of marketers are out there tracking data that they’ll never use is definitely true. But if you can figure out how to efficiently derive insights from it then you can basically never have enough data.
We use tracking tools, and we rely on third-party data too: Clearbit, TowerData, FullContact, and our most recent partnership, Buzzboard, give us huge amounts of data that we can add to what we harvest ourselves. This third-party data enrichment is a major part of what we do for our clients.
Bottom line: If you can measure it, you can score it; and if you can score it, you can use it to score leads. With lead scoring automation, prospects will convert more. You can create a score for anything so people are always being scored and being moved around the system based on their score.
We found one of the most important issues was that we could use quantitative data — numerical scores based on how many times did a prospect do this, or how many times did a customer access that — to answer a qualitative question: what solves this person’s problem? What takes away the pain that they’re feeling right now? (That question is as old as the hills, we’re just using new tools to identify the problem.)
Lead scoring is shaping up as we speak. Please share your own examples with us in the comments below.