7 FAQs About Predictive Lead Scoring with Infer
July 8, 2015
By Paul Schmidt
You've probably heard the term "predicitive lead scoring" floating around and wondered how it impacts you and your marketing. First, you should know that predictive lead scoring is incredible and will change the way you look at qualified leads. Second, HubSpot has integrated with Infer, the leader in predictive lead scoring.
Recently we caught up with Sean Zinsmeister, Director of Product Marketing, for Infer to help you answer the most frequently asked questions:
1. How is predictive lead scoring different than my own custom lead scoring in HubSpot? Would it ever make sense to use them both?
There are two big differences between predictive lead scoring and the kind of custom lead scoring in marketing automation platforms (MAPs). The first is that predictive uses both internal data from your CRM and MAP systems plus thousands of external signals from a variety of data sources outside your company. The second major distinction is that predictive scoring solutions use machine learning to look at all kinds of combinations in the data that humans could never grok on their own. Whereas MAPs require you to manually come up with points-based calculations formed through your gut instincts, predictive solutions take the guesswork out of the equation and do all that work for you in order to better predict higher converting leads.
Depending on your use case, it often makes sense to leverage multiple different flavors of lead scoring. Fit scoring, which looks at a lead’s demographics and a company’s firmographics to find prospects who look like past won opportunities, can give you different insight than behavioral scoring, which looks at a prospect’s intent signals from activity on your web site. Across Infer’s customers, we’ve found that a helpful best practice when first engaging with predictive is to focus on building a really solid fit score that demonstrates success to the business. Once you’ve seen a lift in your average order value and conversion rate, at that point it’s a good idea to consider a two-filter system – which first uses a predictive model to find your best-fit leads, and then uses custom behavioral scoring inside your MAP to pinpoint which of those leads are most likely to buy soon. And after that’s working well, many companies take it a step further to make the second filter less manual by replacing it with a predictive behavioral model.
2. Often times, when we talk about lead scoring, we think about how this is going to benefit the salesperson by allowing them to prioritize/segment leads. How does predictive lead scoring help marketers on a day to day basis?
Predictive lead scoring offers several benefits to marketers, especially as a way to measure the impact of campaigns. For example, it’s a great tool for implementing an agile, “test and invest” inbound marketing strategy. With predictive, you don’t have to wait until the end of a campaign to count up how many leads you generated (which doesn’t really tell you about quality anyway). Instead, you can run two campaigns simultaneously and measure how many high-scoring MQLs are coming in each day. You might use this approach to evaluate marketing activities and regularly optimize your mix of content, syndication channels, or advertising creative. As a result, you can be sure you’re placing your bets on the right channels and driving the most relevant buyers to your sales team.
3. Relationships between sales and marketing can be turbulent at times. For a company to benefit from predictive lead scoring, who should be involved in implementation and rollout of this integration?
Typically, the predictive lead scoring rollout process is marketing-driven as you build out your marketing technology stack. However, it’s very important to involve sales early on as you start to define the objectives and criteria for which leads you’re going to pass to reps. If you do, they’ll have a better understanding of what each predictive score is telling them, and you’ll be able to more easily earn the trust of your reps. Sometimes, the predictive model build process even bubbles up new signals that sales and marketing teams can use for personalizing campaigns, or adjusting and tightening segment criteria. The more both teams are involved in the rollout of predictive, the more you can align effort with impact and double down on the leads that will increase lift.
4. What specific data from Infer would go into scoring my leads? Could I tweak the scoring for these leads if I want to change the scoring criteria (e.g. can I choose which data Infer is predicting on?)
Infer’s models mine companies’ historical customer data from internal systems, and pull in valuable external data points about an individual and the organization they work for – such as relevant job postings, patent filings, social presence, and even the technology vendors they use. While machine learning is our dominant technique for building quality models, part of our process is to develop a great understanding of a new client’s business and their business model at the outset of the discovery process. By doing qualitative research and gathering customer input during the model build, we’re able to incorporate any personas, criteria or data points they’ve identified from the start.
Model performance is everything. This is why we monitor model performance on a constant basis in order to ensure that models that are in a customer’s production process are performing adequately. We periodically refresh these models manually every 3-6 months or as needed in the event that drift occurs between the initial performance and the actual performance.
5. How would companies use Infer to improve/modify their SLA?
Predictive gives you an easy-to-understand architecture for your sales SLAs, as described in detail by Jamie Grenney in a great post on this topic published on the Hubspot blog. Predictive-informed SLAs let your teams lead the conversation with objective data vs. emotion, and they give you more concrete insight than the guessing-and-checking approach of traditional SLAs. Of course, these SLAs are never a replacement for a transparent, ongoing sales and marketing feedback loop, but they certainly help to provide a guiding light throughout the process.
6. What is Infer's competitive advantage in the marketplace?
Infer is unique in the predictive space for many reasons – first an foremost because of our DNA, which is made up of a strong engineering culture arising from Google, Microsoft and Yahoo. We are vicious about acquiring data and finding the areas where data science can unlock the most value for B2B sales and marketing. Because our goal has never been to build a consulting company, we’ve remained laser-focused on model performance and driving impactful results for our customers. That’s why we encourage competitive bake-offs and let our tech and engineering excellence, and model performance do the talking.
7. For those HubSpotters new to lead scoring, how would you recommend they get started using predictive lead scoring?
First off, look at the pain points you’re trying to solve in the business and set goals around that. For example, if you do such great content marketing that you have tons of leads but your sales bandwidth is limited, your goal should be for reps to have the most valuable conversations possible. A great way to narrow in on the best use cases for your company is to talk to other companies who are already seeing success with predictive, and borrow from their playbooks. Learn from those who are finding many different ways to benefit from predictive in both sales and marketing – folks like Zendesk, Box, Zenefits and Hubspot itself.
About the author
Paul Schmidt is a director of client services at SmartBug Media. He works with clients on SEO, analytics, lead generation, sales enablement, customer success and inbound marketing strategy. He previously worked at HubSpot, helping develop inbound strategies for over 200 clients. His past clients include: Travelers Insurance, Unilever, and the SABIAN Cymbal Company. Paul studied percussion in Las Vegas and got his MBA in marketing in Boston Read more articles by Paul Schmidt.