You don’t need a crystal ball to predict whether or not a lead will turn into a new customer. In fact, you already have what it takes to turn your sales team into a lead-closing machine—the insights of your top performers, and your sales activity data.

Your best salespeople run on instinct—most intuitively know which leads are good and which ones are bad. They spend days researching each one of these leads. And they do something else no computer can do (yet)—aggregating and absorbing industry and world news, stock movements, LinkedIn updates, and myriad other sources of information that give them a much better picture of every lead. And they then use that information to prioritize where and how they spend their time. Yes, there are emerging companies that are trying to do that right now. But as of today, the AI is not there yet to fully replace that intuition.

The sweet spot is in combining the intuition of your top performers with the insights from your sales activity data. While there isn’t an AI tool that will do the whole job for your sales team yet, there is a new way to use predictive technology to codify the experience and intuition of your best sales reps and put it to use throughout your organization.

Step 1: Build a comprehensive sales activity dataset

The important word here is “comprehensive”—if activity data on your salespeople has holes in it, the whole endeavor will go bust. That means you need to create a comprehensive dataset with as close to 100% of sales activity captured as possible. We’re going to spend some extra time on this step because if it’s not done right, the rest of the steps will fail.

You can do this by manually by forcing your salespeople to log every single activity they do—all the emails, calls, meetings, etc. This will give you the data, but kill your sales team’s productivity and morale at the same time.

But there is another way. Imagine an inside sales team with a near 100% rate of salespeople manually entering all their activity data in the CRM. That’s a lot of work. Think about it—every salesperson is supposed to do ~100 activities a day. It takes 2 minutes to create a contact in Salesforce and ~3 minutes to log each activity. With 10 new contacts per day and 100 activities, we are looking at each salesperson spending 350 minutes or ~6 hours per day on manual data entry. Which leaves only 2 hours for the salesperson to do his actual work.

Luckily, you can actually automate activity entry into Salesforce. The caveat here is that the system that does this entry for you has to understand the context of every account, deal, lead, and campaign. Without understanding the context, it will not know which account to attach an activity to, as there are multiple variations or similar company names. (Hint: we can help you do this!)

Furthermore, if your CRM is not configured properly—a system that doesn’t understand the context of every lead, campaign, opportunity, account, and contact—it will create more mess than help. Why? Without context, it will start randomly assigning activities to the wrong object, or not assign them at all., on the other hand, understands the context of every deal and is able to capture and sync this data automatically, and intelligently.

Step 2: Segment out your historical top sales performers

What are the KPIs of your top sales performers? For example, the BDRs whose created opportunities have the highest close rate and the activities of AEs who close the most business in your target segment the fastest.

Use your data to map out the KPIs of each of your sales roles—from inside sales to field AEs. Identify the number of touches needed, on average, to close a deal, and how long it takes to get there. And learn how many contacts are tied to the average closed won opportunity.

Step 3: Identify sales decision moments

Identify moments in the data where salespeople had to make a decision that involved their intuition. For example, converting or disqualifying a lead, moving opportunity stages, or identifying champions in a deal. Identify a point in your funnel where the salesperson had to review a group of leads/opportunities/contacts and use their intuition to pick the winners. For example, a moment when a rockstar BDR looked at a dozen MQLs and chose which ones to start with. Sit down with them to understand the decisions they made and why, so that you can identify places of repeatability and/or automation beyond that one team member.

Step 4: Continually analyze and clean your data

Next, you’ll want to gather the maximum amount of data about the leads before and after those sales decision moments. Make sure every lead has as many bullet proof data points as possible. This will give your machine learning the biggest oomph. For example, what did the person do before they became a lead—was it an event they attended, a form they filled out on your website, or a specific term they were searching for? And what happened once they became a lead—what activities correlate to the most opportunities created and closed? What activities indicate potential factors for an opportunity?

But don’t just set it and forget it—you need to continually analyze those activities, and clean and enrich your data. On a quarterly basis, re-evaluate what your best reps are doing, and the activities pre- and post-decision moment. The indicators for a good lead or opportunity will naturally evolve over time, so your predictive model should likewise be dynamic. For your contact data, you can hire someone on a site like Upwork or another freelance marketplace to double check the dataset, or clean your date with a CRM plug in such as Datanyze, ZoomInfo, or D & B Hoovers. While this may sound like unnecessary work or cost, if done right it will benefit you long term, as your data needs to be as clean as possible to not to give you false results.

Step 5: Feed the data into a predictive engine to provide insights

Now it’s time to feed your enriched data into a predictive engine, such as Tableau. If you are building the engine in-house, like some of our data-science savvy customers, you know what to do here. You’ll be able to define, and then slice and dice the data, by the filters most relevant to your business, and set up self-service dashboards for sales and marketing to use. If you don’t have a data science team, talk to a predictive lead scoring vendor like Infer, Everstring, or 6sense about how they can incorporate this data into your predictive models.

Step 6: Give the team access to the predictive engine

While your first instinct might be to limit access to the predictive engine to your sales management team, fight it! By giving your full sales team access to the predictive engine as they are making similar decisions, your predictive model can look at past decisions and provide an intelligent suggested next best step. For example, it may tell the salesperson to set a higher lead score, add a priority tag, or send a nurture email as a next step.

AI takes your predictive lead scoring to the next level

Predictive lead scoring, powered by AI, allows you to leverage your best asset: your top performers. When you use AI as part of your sales process it helps you scale the intuition of your best hunters to the rest of the team. No more trial and error: your sales team gets trained on exactly what works for YOUR sales cycle.

At one of our clients, a large networking and security company, they’re working on taking their predictive lead scoring and predictive analytics even a step further. Their ultimate goal is to move towards “prescriptive analytics,” i.e. being able to deliver a set of actions that employees throughout the company should take to convert individual prospects. They’ve started moving towards this by collecting data on every customer interaction they’ve had, in every channel.

Once you’ve created a single source of truth with your customer data, you can analyze it and start seeing the patterns and data correlations, and move from predicting customer behavior to prescribing actions that should result in closing that sale.

To find out how can supercharge your predictive lead scoring, request a demo.