7 Questions for Those Considering Predictive Lead Scoring (+ Free Checklist!)

7 Questions for Those Considering Predictive Lead Scoring (+ Free Checklist!)

It can be confusing to work out if predictive lead scoring is suitable for your company. In this article, I pose seven question for you to consider before going full steam ahead with predictive lead scoring. Remember though, these are just guidelines!

Don’t forget to check out “What is Predictive Lead Scoring?” if you want some background information first!

As you read down the list, you may come to realize that now is the not the right time to pursue predictive lead scoring. This list will give you a good idea of what you’ll need to be ready to implement it in the future!

If you’d like to skip all the reading and head straight for the downloadable checklist, simple scroll down to the bottom of this page!

#1 Do you have enough leads to start with?

You need data to build a predictive model and more importantly – to build a decent model, you’ll need a decent number of leads. I know, this immediately begs the question, how many is sufficient? Well, the more the better in terms of creating a good model (and by good, we mean accurate). In terms of a minimum, unfortunately there is no hard and fast rule. It really depends on a number of other factors such as the quality of your data (see #3 for more details). Some suggest as few as 100. Others, say you need quite a few more. For example Hubspot’s app mandates “no less than 500, and twice as many should be customers vs. non-customers”.  The latter part of that statement is an important point: you need have data on lost deals as well as successful deals. This information enables the algorithms to figure out their likelihood to purchase or not purchase – all that information feeds into their overall predictive score.

#2 Do you have a steady stream of leads?

A key point of predictive lead scoring is that it aims to help you prioritize leads so you need more leads coming through the door than you want your sales team to being dealing with on an individual basis. That’s one of the reasons why predictive lead scoring is typically used by growing companies or established companies with existing traffic. It’s not appropriate for a start-up that only has a handful of leads trickling in. At start-up phase, you’ll likely be wanting to nurture each lead closely to learn as much as you can about them.

#3 Do you have quality data on the leads?

If you think of the number of leads as length, then you also need to have depth to your data as well. A machine learning algorithm is not going to work well if you’ve got only a handful of variables. You should have explicit information (name, email, location, interests, etc) as well as implicit information (that is, behavioral information such as which pages of your website they’ve visited). You can sometimes further “enrich” the explicit data by linking your contact with external information (e.g. their web behavior, info from their LinkedIn, etc.) so that the machine learning algorithm can best determine if there are patterns of behavior. For example, are there particular pages that prospects frequently visit before converting?

It is the behavioral data that is key for a good predictive model. A lot of traditional lead scoring focuses on information about a lead (who they are). Predictive lead scoring, by contrast, can leverage information on a prospect’ patterns of behavior (what they do). So you want to make sure you can harvest lots of behavioral information. The more the better. There are providers out there who can do that (but for a hefty fee!).

#4 Can you efficiently capture data on your leads?

No point having good data if you can’t efficiently store it and keep your database up-to-date. Data should be complete (i.e. doesn’t have lots of missing values), clean (e.g. structured appropriately), and up-to-date. This is all easier said than done as it requires diligence to maintain good data. If you’re enriching data with external information, you’ll want make sure you can seamlessly integrate the data back into your CRM/database.

Ideally, (though not essential) for predictive lead scoring, you’ll want to be able capture and store information in real-time. One of the major benefits of predictive lead scoring is that the models are adaptive. It changes depending on the data it’s using. This is in contrast to traditional lead scoring which is a static model (and so can quickly become obsolete, especially in a rapidly changing market).

#5 Do you know how to (best) use the scores?

Predictive Lead Scoring is more than just supplying a single figure to sales. There are many use cases where marketing, sales and management can take advantage of predictive lead scoring. This includes lead profiling, segmentation, product cross-selling/up-selling, product upgrading, automated marketing, validation and/or realignment of business strategy. Also, predictive lead scoring can provide valuable information along every step of the sales conversion funnel (helping you nurture the leads). So being clear about what predictive lead scoring can do for you, and what business problems it will be solving, is vital for your success in implementing it.

#6 Have you got the resources to implement predictive lead scoring?

Predictive lead scoring takes some effort to implement. As discussed, you need to gather data, clean it, manage it, enrich it, audit it, change it, etc. That all requires some commitment and cost. Especially if you’re using a vendor to enrich your data and potentially use their in-built predictive lead scoring systems. Some CRM’s offer a full service (data storing, enrichment, modelling, score implementation). That might be fine for some, but particularly for a growing enterprise – the price tag associated with these vendors may be tough to swallow. This is where Displayr can help).

For effective implementation, predictive lead scoring also requires an alignment of marketing and sales. The teams need to work together to provide the scores, use the scores, and provide integrated feedback. For some companies, where the teams operate separately, this can throw up some challenges. But equally, the implementation of predictive lead scoring can promote cross-team harmony and integration.

#7 Do you have clear expectations and measurable goals?

All of the above points highlight the time, effort, and money that needs to go into adopting predictive lead scoring. But like anything worth doing, it’s an investment. And with any investment you should be clear about how you are going to measure your return. The obvious measure is an increase in sales conversion rate, but you could have other measures as well depending on the business problem(s) predictive lead scoring is addressing. You should also have a view about what timescales are realistic and desirable. Many users of predictive lead scoring report great results in as little as a few months.

How we can help with your predictive lead scoring!

At Displayr, we’re providing an easy-to-use DIY tool for business intelligence. Part of its comprehensive analytics toolkit is a suite of machine learning techniques that can be used for predictive lead scoring (we’ll be covering this in future articles, so stay tuned!). So rather than using vendors to do the magic, Displayr gives you a more cost-effective solution by bringing your predictive lead scoring in house!

About Matt Steele

Matt has over 14 years of experience in the marketing research arena, with a combination of research experience (qualitative and quantitative), marketing training, academic psychology (cognitive), creative leadership, geekiness and artistic flair. He currently works for Displayr (the home of Q and Displayr) and is based in London: supporting, selling, marketing and training for Q research software and associated software packages (eg: Displayr). He holds a Honours degree in Psychology from UNSW, a Grad Cert. in Marketing from UTS, and a Grad Dip in Directing from NIDA (all based in Sydney, Australia).