Let me start with a confession. Data, analytics, algorithms, statistics, probability and all such left-brain terms make my head spin. If you are a right-brained marketer like me, then you prefer to go by gut and instinct, love conceptual explanations and enjoy creating solutions rather than having a machine give them to you. Unfortunately, anyone remotely connected with marketing today, cannot escape data, numbers, fact and technology. Analytics is all of that and more. It’s time to face (and slay) our fears. This article will attempt to decode Predictive Marketing Analytics so it makes sense to a B2B marketer thinking about the what, why, when and how of it.
Why analytics? Quite simply, it leads to better decisions. Informed decisions. Data-driven decisions. It connects various data points and sets to identify trends, patterns, anomalies and gaps. Luckily, the technology available today, does not do just that, but also presents the findings in a way that a regular (right-brained) marketer like you and me can make sense of and apply the insights and decisions to marketing action plans and execution strategies.
The basic concept is simple enough. You have data –> you use technology to run some analytics on it –> you get some actionable insight from it which will help you make better decisions –> that can drive better marketing results & ROI.
So, what’s the catch? Well, there are a lot of different sources of data, there are lots of different kinds of analytics (SEE BOX: Types of Analytics) and there are several use cases for which you can deploy decision science solutions.
Â
Â
With limited resources to invest in technology, and increasing pressure to show results from those investments, marketers need to be sure they are going in the right direction and not just the most popular one. So, let’s zoom out for a bit and put this in perspective.
What does the analytics ecosystem look like? – Big Data, DMP, CDP, AI, CRM, Analytics, Marketing Automation, Intelligence, Machine Learning: all play starring roles but how are they related? This chart attempts to explain the relation and flow of events between the star players in the larger ecosystem.
Â
Â
Chart 1: THE ANALYTICS ECOSYSTEM BIG PICTURE
Opens a new window
So now you have an idea of the bigger picture and where and how the major pieces fit, it’s time to zero in on the question du jour: the what, why, when and how of Predictive Marketing Analytics.
The age of predictive marketing is upon us. It is no longer enough to just track historical customer behavior and then tailor our response based on that behavior. Now, it is possible, and perhaps even critical to predict future behavior based on historical and real-time actions; to anticipate, and proactively address upcoming customer needs and wants. To paraphrase Jack Welch, it is not enough to be better than your competitor. You have to set a whole new (and hopefully unique) standard in customer experience. To do this, it’s important to know your customer better than anyone else. What they have done, what they are doing, and what they will do. Going a step further, perhaps it is even about proactively nudging customer behavior in certain directions based on what we know about how they have acted and may act in the future. Predictive Marketing Analytics can help do all of that.
Chart 2: Let’s zero in on predictive analytics for (B2B) marketing in the larger Analytics ecosystem.
Opens a new window
How can you specifically use predictive marketing analytics today?
Predictive Marketing Analytics has several applications (or what vendors tend to call ‘use cases’) for B2B marketers. Let’s look at the situations in which it may come to your rescue.
Opens a new window PROSPECTING (Demand Generation with lead enrichment and scoring)
- Data diagnostics: If your first-party data is in bad shape, disconnected, incomplete, inaccurate, obsolete, then the predictive analytics vendor will help you bring it to a level of acceptability, wherein it can be used as input data for analytics.
- Data quality: There are bound to be inhouse data silos across functions and platforms. Externally, several disparate sources of data exist, from browsing behavior to government records. Unstructured data (voice transcripts, emails, chat records etc) pose a whole different challenge. Predictive Analytics technology can help connect all of the disparate data and make each record more complete and detailed by matching and populating it from multiple sources. This, in turn, helps analytics to build more robust, holistic ‘personas‘, segments and journey stages – by sorting through and connecting thousands of characteristics and signals to determine which are actually good indicators of future behavior.
- Identify the best prospects (within a segment) most likely to buy from you (based on matching characteristics with existing customers – aptly called ‘lookalikes‘) and invest your resources in reaching them rather than trying to reach an entire segment.
- Expand the prospect base: Using the ‘lookalike‘ profile to identify prospects not currently in the marketer’s own database, by trawling through thousands of data points in external data sources.
CONVERSION (build pipeline and tailor response to leads based on their buying intent score and probability to convert)
- Prioritize the leads you already have, and focus on those most likely to convert in the immediate future based on lead scores and likelihood to convert predictive models
- Reinstate cold leads by finding other contacts within the same organization that can be reached by Sales teams
- Lead routing can route leads scored on their buying intent to either Sales teams or to the right (nurture) campaign on your marketing automation platform
- Account-based marketing: Profile and identify the right roles within the right prospects for better ABM success
- Predict success probability of campaigns targeted at audiences picked via predictive segmentation
RETENTION (keep and grow existing customers)
- Manage and pre-empt churn by identifying churn-prone behavior well before renewal dates
- Cross-selling and upselling to existing customers by offering the best-fit products and solutions at the right time to existing customers based on the probability of conversion (Note: All vendors may not provide solutions for all use cases, and some may offer more than this)
Is your B2B a viable candidate for predictive marketing analytics?
Predictive analytics is not for everyone. It is not so much about size as it is about maturity though. B2Bs small and large, at a fairly advanced stage of marketing technology in terms of having both, a stable CRM and a marketing automation or ABM stack up and running are ideal candidates for predictive, because leveraging predictive insights can exponentially improve ROI from those installed marketing technologies. As John Hurley, Head of Demand Generation at Radius said in a conversation about predictive,
MA and CRM are workflow tools, predictive has all the data and rich analysis on who should be in each marketing tool
But there’s an elephant in the room
Even though the quantum of data generated has increased exorbitantly, and data is the bedrock of analytics and informed decision-making, analytics is not, ironically about more data but about making better sense of the data available, albeit at scale.
So, let’s say you have decided you need predictive and you also meet the basic criteria for deploying it. You call on several vendors and in every meeting, there is an elephant in the room that no one wants to deal with… no, it’s not your data, it’s the quality of your data.
While having a stable CRM and MAP or ABM platform in place is a prerequisite to deploying advanced analytics, if the quality of data inside those systems is not so great, I don’t have to spell out what the results of predictive will be. The data has to be usable, current and actionable. It’s a sensitive topic because no one wants to admit that their million-dollar system is full of junk. Or, blame anyone who is responsible for putting the junk in there (especially if they are themselves to blame).  Another reason could be that CRM is typically a sales-owned system ,and this evaluation is being led by marketing, who may not want to get confrontational about it and are also under pressure to deliver results. If this was FB, I’d be really tempted to say, “It’s complicatedâ€. Oops, I just did.
Â
Â
Â
And if you find (or admit) that your data is lousy, then how do you move beyond that?
Bad data is costing you more than you can imagine, in terms of opportunity and revenue. Luckily for you, it is in the interest of the predictive analytics vendors to help B2Bs get their data sorted to a somewhat acceptable level. Without this, their category is not going to grow. So, the good news is that every (good) predictive analytics vendor will most likely do the below for you:
- Offer you access to a third-party owned or licensed data set that you can get started with
- Help you, through a process known as ‘data stewardship’ which includes, CRM (data) diagnostics, cleansing (deduplicate, remove inaccurate, incomplete and obsolete records etc.), and enrichment (fill in gaps, add fields to existing records from external sources etc
- ‘Match’ their third-party data with your data and look for a high match rate. A good match means higher probability that you will get the high-quality net new leads or improved database richness that your aim for.
Opens a new window But it doesn’t end there. Even if your CRM or MAP is decent in terms of primary records, here’s another bummer: the data in the fields of those records will ‘decay’ faster than fruits in summer. Many vendors provide access to high quality third-party licensed databases, and once you are subscribed, will match and update / append your existing records constantly to ensure the data fields stay fresh and relevant over time.
Â
Â
You can run but you can’t hide: the future of predictive marketing analytics
B2B buyers are being conditioned by their B2C experiences, and expect the same personalized contact and quick turnaround that B2C offers. Big, complex data with complex, dynamic relationships and interconnections; where marketers need to respond to changes instantaneously, will need predictive analytics to manage it at some point. It’s not a question of if, it’s when.
As Mintigo puts itOpens a new window , “The state where predictive modeling drives core marketing processes, can be called predictive marketingâ€
Assuming most B2B marketers will strive towards making increasingly robust data-driven decisions, the rise of predictive marketing analytics seems inevitable.
In spite of this, and despite what the surveys say, real-life practitioners and vendors tell me that there is still resistance to committing fully to predictive analytics today.
So, apart from the fact that overly logical and algorithmic discussions make our heads explode, what holds back B2Bs (that meet the criteria) from embracing predictive analytics as a strategic marketing tool?
- Maintain the status quo: That hoary old chestnut surfaces again. It is an argument I hear across all categories of technology. CMOs don’t see the urgency in it and don’t prioritize it. Unfortunately, the world is moving to a Data Economy. Everyone – especially your prospects and customers – expect you to be coming from a place of in-depth insight about what they want and need. Ignore that reality at your peril.
- Make or buy: Many B2Bs think they can do a better job of building the technology themselves. I had referred to this in my article on Key Account Management TechnologyOpens a new window a few months ago, and I will say it again: this approach is like waiting for the meteor to come and hit you. In today’s world, when cloud-based SaaS technologies offer full-feature specialist solutions to address complex problems at the cost of one month’s subscription, it is not viable to go down the ‘invest in non-core areas’ strategy.
- Wait and watch (while using existing tools): If you are thinking ‘Hey! I just invested loads of time and effort getting users to adopt Marketing Automation and now there is this new thing in between my data and my execution piece’ you wouldn’t be alone. Since most marketers have already got some kind of descriptive analytics in place (BI, dashboards, intel), they may be quite happy (and not wrong) in thinking it is enough to move forward with execution based on that sort of static intel. It is important though to remember that most static BI or descriptive analytics is based on the assumption – ‘what happened in the past will happen in the future in the same way’. However, the Data Economy demands a more forward-focused approach to managing marketing data. Your prospects and customers are evolving rapidly. Your data analytics and insight generation machinery needs to be as dynamic as them. Or, else your intel and reports could be just as useful as yesterday’s news.
- Go for a cheaper option: And, that is typically buying data, records, lists from third-party list vendors. It may be a decent short-term option, but if you are going to an external data source, you already understand and appreciate that the data challenge needs a sustainable, long-term and strategic solution.
- Change is scary: You have seen others try it and fail. Your own people will resist change, because they already have too much on their plate. You are still smarting from some other post-purchase dissonance issues. All valid. But, the crux with investing in this or any technology is to know what you are going to solve before you make the purchase. This is not something to buy, because it sounds sexy and everyone is talking about it obviously, but it’s also not something to buy if you don’t really have a valid use case for it. Work closely with the vendors pitching for your business to identify clear use cases and then decide if it’s worth the fight.
That said, as it stands today it’s easy to see predictive analytics marketing as just advanced prospecting or segmentation technology. It doesn’t help that vendors describe their services in many different ways – from audience management platforms, AI for business, AI assisted sales & marketing, predictive intelligence platforms among many other descriptions.
While indeed the dominant use cases today are data stewardship, data enrichment, predictive segmentation and lead scoring; the core offerings will evolve as the vendors and users mature. It is important to remember though that investments in data quality and data partnerships today, will not be wasted and can only serve to improve results from more complex predictive analytics moving forward.
Ultimately, the work of all technology, strategy and innovation – no matter how fancy they sound – is to drive pipeline and conversions. And, that is something predictive analytics will only get better at. Some specific areas of maturity are:
- Predictive analytics will continue to get better at pulling many different sources of data – and connect thousands of characteristics, variables and behaviors together – to deliver ever more of the perfect customer profiles, micro-segments, intent to buy and probability for success models it is designed to generate.
- The 3rd party databases offered by predictive analytics vendors today will look very different from the dominant data licensing model we see today. Already Radius has the Network of Record, Everstring has its Company Graph, Microsoft acquired LinkedIn (thus giving it access to the wealth of B2B data it owns) among others. IBM Watson offers the UBX (Universal Behaviour Exchange) –  a cloud-based offering that lets you combine data generated from many integrated IBM and Business Partner ecosystem solutions to deliver a complete picture of your customers’ behavior so that you can act faster and create more personalized messages to improve engagement. Vendors are investing big to develop purposeful B2B databases that can be used to take B2B predictive analytics to the next level. Along with inhouse data also getting smarter and better & machine learning getting more advanced; high ‘data accuracy’ levels are going to drive much more accurate predictive intelligence. It will be an era of strategic partnerships between B2Bs and their analytics vendor.
- Intelligent platforms of the future, armed with deep and powerful databases as described above, will not just be able to identify the best-fit prospects or traits most likely to convert, they will be able to proactively and pre-emptively recommend specific named prospects and customers to the Sales teams at the right time for the right portfolio offering.
- This will come with tighter integration between the three core components of predictive analytics – data, analytics and execution components. In other words, the backward integration of analytics with Data and the forward integration with execution platforms like CRM and MA will only make the recommendations more robust, actionable and real time.
Right-brained marketers, don’t say we didn’t warn you! Predictive analytics – like math and physics – seems like it’s here to stay. But hopefully, you are not that intimidated by it anymore. Good luck!