Predictive Analytics is the Future of B2B Digital Marketing


Share on LinkedIn

Over the past 10 years or more, B2B marketing has essentially been defined by process automation, including lead generation, nurturing, email campaigns, and so on. The big driver: buyers are increasingly turning to digital channels for self education and, let’s be honest, to avoid talking to those pesky sales reps.

Does this mean RIP, sales professionals? Yes and no.

  • Yes, if sales reps continue to make calls unprepared to add value. Asking lame questions like: “What do you do?” or “What keeps you up a night?” Most of the prospecting calls and emails I receive show no effort whatsoever to find out what I do. It’s all out in the open on CustomerThink, LinkedIn and other sites where I post.
  • No, if sales reps “up their game” to make calls worthwhile. Some researchers claim that 60% or more of the buying process is now being completed before a sales rep is contacted. That may be true in the aggregate, but it doesn’t mean Sellers should just wait for prospects to call.

Gone are the days where reps are the main source of product information. Buyers can search online, ask questions in online communities, or network privately with their peers. Some of these public activities are buying signals, but who has the time to do that detective work? Not inside sales reps with a dialing-for-dollars daily quota, that’s for sure.

Personal engagement still matters

But it would be wrong to assume that just waiting for the phone to ring is being buyer-centric. A recent Gartner study by Hank Barnes and Tiffani Bova concluded that, for IT buyers:

Personal interactions with providers are still the most influential activity in B2B buying decisions. However, buyers do not value their interactions with salespeople as much as they did in the past. Sales must adjust processes and skills to learn to guide buyers through their purchase cycle.

According to their research, “Direct interaction with the provider” was rated the most influential marketing activity, scoring 5.5 on a scale of 1 to 7. A close second: customer references (5.3). White papers, events and presentations ranked lower.

However, the study didn’t conclude it should be the sales rep that makes the engagement. In fact, respondents ranked interactions with industry and technical experts much more highly, regardless of whether they were in the early, mid or late stage of buying. That’s still not good news for “lone ranger” reps that aren’t adept at orchestrating company resources.

(By the way, Barnes has developed a really interesting way of looking at buying phases, which he writes about at Teams, Streams, and Provider Dreams.)

Barnes said that sales models and channels are under stress, but they’re “not seeing a lot of change yet.” That’s consistent with what I’ve been seeing the past few years. Lots of hand wringing and discussion, but sellers are struggling to change. Mainly because, according to Barnes, sales organization are still under the gun to make their quarterly numbers. I’d equate this to a football team that has developed a certain game plan, and can’t change in the 4th quarter even if it’s losing.

Complexity is the Achilles’ heel

OK, so sales organizations are struggling. Is marketing automation the answer?

Clearly there is a growing educational and nurturing role for marketing. Over the past 10 years the industry has responded with robust solutions from Eloqua, Marketo, Hubspot and dozens more. They claim that by “scoring” leads based on demographics (job title, industry, etc.) and activity (clicking on white paper, navigating web pages, etc.) that marketers can deliver better “marketing qualified leads” to sales.

Well, it’s certainly an improvement over dumping all leads on sales reps. But earlier this year, I lamented in “What’s next for the Marketing Automation industry?” that complexity is the Achilles’ heel of the MA industry:

… marketing automation solutions require marketers to think like programmers. I get a headache thinking about what marketers have to deal with now — personas, content, campaigns and lead scoring — and I’m an analytic sort with a math degree and some programming experience.

Campaign workflows are at best a judgement call, because few organization do a quality job developing buyer personas and content strategies by stage and persona. Further, the all-important lead score is basically an intelligent guess. Here’s how the scoring algorithm is built in most organizations: “OK, let’s give 5 points if someone is a VP, 10 point if a CEO, 4 points if they already have, and 20 points if they say they’re ready to buy.”

Better than nothing, but what if the Sales Operations manager is actually more likely to drive the buying decision, and not some upper level muckety muck? What if the most telling buying signals are not under the Seller’s direct control?

Predicting order from chaos

Making sense out of complexity is the perfect job for Predictive Analytics (PA), which I believe will be savior of the MA industry. Instead of guessing about lead scores, a predictive model can be built which correlates lead characteristics (demographics, behavior, etc.) with business outcomes (sales qualified leads, conversions, long-term revenue). And isn’t that the point of scoring?

Recently I was briefed by two companies doing some pioneering work in this area.

  • Leadspace is a Battery Ventures-backed startup launched in 2007, and focused on B2B sales marketing since 2010. The company was co-founded by Amnon Misher in Israel and is now expanding fast into the US market. The idea: use analytics as a kind of Private Investigator to find buying signals on the Social Web. This is truly like finding a needle in a haystack, because B2B buyers are limited in number (compared to consumers) and engage less often publicly. Social signals are combined with CRM, MA and other internal company data to build a predictive score.

    Leadspace has a “matching” technology that uses the prospects email, name, company (typically provided via a MA form) to feed a “lead targeting engine” that identifies the leads worth reaching out to. Social signals that might increase a lead score might include following a competitor’s Twitter account, posting a question in public forum, public interviews and more, says Misher.

    The key point is that the score is predictive–built using the company’s own data. Not some kind of guess by a frazzled marketer trying to get the next campaign out before sales screams for more leads.

  • A similar approach is offered by Infer, a VC-backed startup based in Silicon Valley. Like Leadspace, Infer mines “big data” in the public Web to find buying signals, then combines with internal data and builds a predictive lead scoring model.

    Co-founder/CEO Vik Singh says (and I agree 100%), “the answer is not every organization hiring a data scientist.” This is exactly the point I made in my earlier blog post: systems need to figure this stuff out, not people. Hiring statistics geeks or expecting MA professionals to learn to be data scientists is a non-starter for all but the largest and most progressive organizations.

    Infer-based models are personalized to each company using their own data plus public information, including information from data brokers. Singh says that they can build a model and start pushing lead scores within days of getting access to CRM data. Infer started shipping in early 2012 and has early customers in the tech sector.

There are other companies active in the B2B “predictive” space, including Insightera, Lattice, and ReachForce. I predict that it’s just a matter of time before most of these innovative vendors get acquired by the big players in B2B marketing and sales, such as Oracle,, or the newly public Marketo.

Leveraging predictive analytics, B2B marketing/sales will be much better prepared to deliver a win-win buying experience. Companies get help deciding which prospects to engage with. Buyers get fewer bad sales experiences. While it doesn’t solve all of the sales reps challenges, predictive analytics can help focus attention where it’s most likely to be welcome, and pay off. That’s a start.

Further reading:

Disclosure: This post is part on my independent coverage of technology industry developments. No endorsement is implied for any vendors mentioned in this post. Some vendors mentioned have been CustomerThink sponsors within the past year. Please visit our sponsor page for information on companies that have supported this community.


  1. Bob: Though I agree there are great opportunities for predictive analytics in digital marketing, I think prospective buyers should view vendor claims with skepticism. Along with predictions of the ‘next great thing’ comes a significant amount of hype, especially from vendors.

    True that having some predictive capabilities can be better than none, but you have pointed out two great questions: “what if the Sales Operations manager is actually more likely to drive the buying decision, and not some upper level muckety muck? What if the most telling buying signals are not under the Seller’s direct control?”

    I have seen managers place high confidence in results from predictive models when those results are unproven in separating signal from noise. The outcome from that mistake can be as disastrous as it is expensive. One of the biggest revenue risks companies face results from assuming their predictive models work better than they actually do–or can.

  2. Andy, any model, predictive or otherwise, should be validated.

    I think there is, as usual, a lot of hype in analytics, big data, and so on. Lots of solutions trying to find problems to solve.

    In this case, as I’ve been researching common problems in marketing/sales over the past few years, it’s clear that:
    1. sales reps aren’t changing very much at all, despite all the new tools
    2. buyers are using other channels, which leaves sales reps potentially as the only one without a chair when the music stops playing
    3. companies need to make better decisions on who to contact, and how to engage more productively

    Predictive analytics can help. To be useful, it doesn’t have to be (and won’t be) perfect in identifying good prospects. It just has to be better than the current process, which is built on a shaky foundation of marketing judgements about lead scores.

    Analytics is not a replacement for human judgement. For more on this: Five Big Ideas to Profit from Analytics and Big Data


Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here