Bet You Didn’t See These Coming…5 Out-of-the-Box Use Cases for Predictive

Someone told me years ago that you know you’re fluent in a language when you can dream in it, because your brain no longer needs to translate between your mother tongue and new language in order to complete the process. With that as a benchmark, it’s fairly telling that some of my proudest work with predictive (both as a customer and now running Customer Experience at Mintigo) roots into my sleeping hours. Predictive models built to answer myriad business questions, called on by triggers in Marketo, with leads and accounts then enriched and passed to Salesforce for my SDR team to action…it’s become my second language.

More common use cases you’re almost certainly familiar with – prioritization of inbound leads for sales, identification of the ideal customer profile (ICP) that defines your best ABM accounts, and the like. But to get the most out of predictive, you’ll want to think outside the box…backward from what problems you are facing in your business, not only forward from what you believe predictive is meant or able to do.

Whether you’re an existing predictive practitioner looking to get more out of your stack or new to predictive, evaluating whether the investment will be realized in return, here are a handful of creative (and fairly simple to implement) ways my customers and I have pushed the edges of predictive.

Exposing Content Gaps

There are countless benefits that come from identifying your ideal customer profile (ICP), but one often overlooked is getting a better understanding of the strength of your content library to target these folks. Take a look at the Customer DNA that makes up your ICP (not just firmographics, but also job titles, hiring trends, components of your customers’ tech stack, third-party web search topics, etc.) and use these to inform what content you create and how to build your messages. Take it a step further and actually create nurture streams to hone in on relevant topic areas.

Some of this exercise will be super intuitive, and some less so. I used to work for a company that built web server software, with specific strength in handling high-concurrency (sites that get hit with tens of thousands of users at a time). When we saw that the presence of Marketo and e-commerce solutions in a prospect’s tech stack was highly correlated with propensity to buy from us, my SDRs were skeptical, but as we interrogated the data, it was clear that these companies were using their website for demand generation and online purchases, so you bet they cared if it ran slowly or fell over! Starting a call with “I see you have Marketo so you should buy our web server” would have gone nowhere, but “a lot of our customers who use their website for demand generation and online purchasing buy our solution to ensure maximum speed and stability to give customers and excellent user experience” was highly relevant and extremely effective in securing meetings.

Channel & Campaign Analysis

There’s an incredible amount of scrutiny on marketing spend, and since no vendor is going to approach you with “hey, I’ve got a mediocre offer that may or may not be worth your while,” it’s on you to determine what’s really best for your business. The trouble is that with a lot of investments, there is no “try and buy” option and not all “leads” are created equal. A simple example of this is trade shows – if you’re giving away the latest gadget (mea culpa, my girls have a non-trivial number of fidget spinners, many of which came from vendors I will never actually purchase from) you may end up with “leads” in your database that will never convert. This is a huge challenge for companies with open source pedigree, too, or those who create excellent thought leadership content; there is tremendous value to many people to be in your database, but they have little or no likelihood of ever converting, so are of very limited value to you.

Whether after or before investing, use predictive to get a leading indicator on the quality of an opportunity.

For campaigns already executed, score and enrich the campaign members to see how they score or rank. If you are only passing A and B-ranked leads to sales, then a campaign that generates 100 of these is actually stronger than a campaign with 500 members of which only 50 rank A or B and 450 are just noise in your system. Cost per lead is interesting, but cost per MQL-able lead is more so. This can also be done at a channel-level, for example, you may find that tradeshows are expensive to sponsor, but generate very high-quality engagements, whereas webinars have more tire kicker attendees.

Likewise, if you are considering a future investment (vs. analyzing past), see if the vendor can provide you with either a blind lead list (last year’s event attendees, for example) or if you can run a short-term pilot with a guarantee to make a larger purchase if the quality is high. I had a content syndication vendor once try to explain that the reason my leads from them were not converting (yet) was because they were top-of-funnel, where the real issue was simply that they were low ranked in propensity to convert and no amount of time was going to shake revenue from that tree.

Knowing Who Not to Call

Knowing who was the highest propensity to purchase from you is key, but knowing who not to call is can be of equal or greater value. You may have a ton of leads, but if only a small portion are the “right” leads, the total volume is irrelevant and your sales team’s efficiency (note, they’re often amongst the highest paid folks in the company) is abysmal.

As a marketing and predictive practitioner, I worked with my SDR team on a wild and bold test a few years ago – we passed them only 50% of the volume of leads they’d previously been working and asked that they spend twice as much time on each of them by personalizing messaging, getting creative in their outreach strategy, etc. (and comped them on compliance with the process, so they were incented to participate instead of deviate). After just three months, we were able to achieve 85% of our prior conversions out of just 50% of the volume! We promoted the top third of the SDR team (by performance) to outbound/BDR roles and focused them on the top accounts we had not yet engaged with (identified by predictive, also) to try to build those relationships, while the remaining SDRs continued to focus on the very best inbound leads. Within another quarter, we’d blown out our previous numbers. Marketing + sales = win!

Supporting Cross-Sell and Upsell

If you’ve got a single product model in place, another great next step is creating a multi-product, cross-sell, or upsell model.

In implementation, you could score a new lead or account against their propensity to not just purchase from your company, but how likely they are to purchase each of your specific products and customize their nurture or sale messaging experience based on that. For existing customers, you could also use this model to determine what the “next best offer” might be or when there are triggers that would warrant a discussion re: expanding their current contract. If a customer has a legitimate need and you are serving them well in your existing engagement, your outreach will feel more like helping than selling and everyone comes out ahead.

Identify Customers at Risk of Churn or Ready for Upsell

While Mintigo is not formally a Customer Success solution (there are some great ones out there), our models and data have proven invaluable with many of our customers in customer retention and upsell. There are countless ways to slice and dice this, but a few core examples:

With an API integration into your usage or purchase history data, models can look for patterns not just in fit, intent, and marketing engagement behavior, but trends such as spurts or declines in usage, clusters of purchases, and more. Someone who uses/purchases consistently, though seasonally, may not actually be a churn risk if they’re not active at a given moment in time, although someone who is continuing to use/purchase regularly, but at a lower rate, may actually be at risk.

Tech stack information can be an invaluable tool in evaluating customer health. If a customer has added a complimentary solution to their stack, you may be able to work with them to become even stickier by integrating into their larger strategy. On the flipside, the presence of a competitive solution to their stack that you are not aware of, they may be doing an evaluation that you want to reach out about.

And intent data – oh, how we love thee. At Mintigo we get a raw firehose of billions of visits to thousands of third-party publisher sites and use this to create a baseline normal for research on each tracked topic at each company; this ensures that “intent” doesn’t just mean the company is really large, so the volume is significant, but rather we’re looking for a sustained spike in search activity on each topic across multiple users and multiple days. Knowing that an existing customer is looking at competitive offering or is in-market for something you offer and have not sold them yet can be gold for your CS team.

Now What?

Crawling seemed great until I learned to walk, and running was awesome until I learned to fly. It is my goal every day to make my customers look like heroes both internally and with their own customers, so if you’re pushing the edges on this stuff, too, or interested in digging deeper into the architecture of these types of programs, please reach out. I’d love to chat more.


Erin Peterson


Erin Peterson is the VP Customer Experience at Mintigo. She is an independently-driven business professional with experience expediting critical programs and meeting key deliverables for global, multi-billion dollar enterprises, early-stage startups and growth-stage companies. Skillfully manage on-site and remote employees, as well as international virtual teams across multiple business units, while forging valuable partnerships with clients and business associates. Meticulous project manager and facilitator, with demonstrated success in executing large-scale programs and implementing process improvements. Multi-faceted industry experience spans software, hardware, communications, legal, education, and humanitarian organizations.