- Trial-and-error methods for optimizing marketing tactics and results are hampered by difficulties in controlling and linking variables
- Predictive analytics provides marketers with the ability to simulate rather than experiment, and to make predictions rather than guesses
- Eight use cases for predictive analytics can help b-to-b marketers make decisions across the SiriusDecisions Demand Waterfall® and customer lifecycle
Contemporary folklore has it that Thomas Edison conducted 10,000 different experiments before alighting on a functioning electric bulb. More sober estimates suggest a trial-and-error figure in the hundreds for the light bulb, but somewhere in the thousands for another of his great inventions, the alkaline battery.
For most b-to-b marketers, the world still looks much like it did for Edison – nearly every program is an experiment. Marketers face many challenges in linking variables (e.g. audiences, messages, tactics) to results, which hampers the success of trial-and-error experimentation within campaigns. Though still early in its b-to-b adoption, predictive analytics already provides marketers with the ability to simulate rather than experiment, and to make predictions rather than guesses. In this post, we describe the eight most important applications of predictive analytics in b-to-b today, shown in the order in which they occur in the demand creation process. This begins before any demand creation can occur, as organizations size and segment their addressable market, and follows through to scoring customers for retention, cross-sell and upsell.
Use cases are defined for each of the examples below:
- Smart Segmentation
- Predictive Personas
- Tactic Matching
- Prospect Prioritization
- Contact Engagement
- Prospect Sourcing
- Opportunity Scoring
- Customer Scoring
Many b-to-b organizations still use traditional segmentation methodologies based on static criteria like employee count and industry, resulting in a proverbial net that is too big or too small. Smart segmentation powered by predictive analytics introduces a new and more accurate way to organize the total addressable market (TAM), by incorporating rapidly changing market conditions and subtle prospect attributes, such as the ratio of engineers to revenue or quarterly hiring trends. By adopting the new methodology, marketing, sales and product leaders can quickly determine distinct segments to drive focus, personalization and prioritized investment across the business.
Use smart segmentation to answer important questions, such as:
- Are there groups of prospects within the TAM that represent better prospects than others?
- How do I distinguish high-propensity targets?
- Have we identified all viable accounts and contacts, or have we missed some?
- Is our product aligned to the needs of the addressable market?
- Does messaging address customer needs?
What does an ideal prospect look like? Which combination of attributes and behaviors indicates which prospects are most likely to convert? Until now, persona creators had only limited data sources (sales force automation systems, tribal knowledge, buyer interviews) and small sample sizes on which to base their answers, resulting in personas that were indistinguishable from those that a competitor might create. Now, predictive personas make it possible to collect statistically valid samples of data that describe a larger variety of attributes and behaviors (e.g. influencers, what media buyers prefer to consume, the Web sites they visit, the social media groups they belong to).
Use predictive personas to answer important questions, such as:
- Which personas are most likely to respond within our best-fit segments?
- What can we say about our buyers to distinguish them from buyers who are more likely to go with a competitor?
- Which attributes and behaviors define the different roles buyers play in the purchase process?
- What weighting should be applied to each of the attributes that comprise our buyer personas?
One of demand creation’s primary responsibilities is to facilitate buying processes by communicating with buyers using relevant messages delivered via the most influential media. Marketing automation’s answer to helping demand creators with this daunting task is lead nurturing, which uses “if, then” logic to deliver sequenced content to prospects based on their last actions. While nurture is still a critical component of the b-to-b marketer’s arsenal, there is a new generation of tactic matching enabled by predictive analytics, which incorporates signals that occur outside of lead nurture flows. With tactic matching, prospects get just what they need to know when they need to know it.
Use tactic matching to answer important questions, such as:
- How do I match tactics to segments and personas to effectively attract and convert inquiries through the SiriusDecisions Demand Waterfall®?
- Do our nurture programs and messaging address buyer needs? Should this messaging be adjusted?
- How can I automatically detect where buyers are in the buyer’s journey and determine the best next offer?
Along with nurture, marketing automation gives marketers the ability to score leads based on attributes and behaviors, with the ultimate goal of identifying which prospects are most likely to convert. In its traditional form, lead scoring asks humans to do a task that computers are better qualified for – machine learning. Prospect prioritization fixes this problem by identifying what are often obscure patterns in vast amounts of unstructured data (e.g. digital signatures that indicate which prospects use compatible technologies), testing endless combinations of attributes and behaviors, and incorporating digital signals that are invisible to marketing automation. Predictive analytics also assesses leads based on account-level behavior so a prospect company with eight people downloading your assets (e.g. white papers) is prioritized above those with just one person downloading them.
Use prospect prioritization to important answer questions, such as:
- Which leads are most likely to convert into customers?
- Is my lead scoring too porous or too stringent?
- Should I be scoring at the account level rather than the contact level, or at both levels?
If you ask lead development representatives (LDRs) whether they should apply the same amount of effort to all leads, the answer will invariably be no. A much better strategy is to spend more time on the leads that have the highest propensity to convert. Contact engagement made possible through predictive analytics removes the guesswork and determines how many attempts should be made to reach a given prospect, along with what days and times are best for these contact attempts. By implementing the contact engagement application, marketers can expect a lift in conversion rates through telequalification, plus a larger volume of more highly qualified leads for sales to follow up on.
Use contact engagement to answer important questions, such as:
- Which leads should our LDR team apply the most effort to?
- What should the volume and cadence be for reaching out to a given prospect?
- When should call attempts be made for a given prospect?
Inbound marketing rarely attracts and converts all the quality buyers a b-to-b organization needs to fill its pipeline. To address the shortage, prospect sourcing examines historical sales results combined with a broad array of third-party behavioral (e.g. buying research, intent), firmagraphic and demographic data to develop a lookalike model – against which potential prospects can be matched and prioritized for acquisition. When models include buyer intent data, marketers can source prospects that are a good fit, along with those who are likely to be in-market for a solution right now.
Use prospect sourcing to answer important questions, such as:
- Are we able to identify all viable accounts and contacts?
- Which accounts that are not in my current marketing or sales databases should we be calling on?
- Which non-responder accounts are likely to be in-market for solutions like ours?
A pervasive problem in b-to-b organizations is lack of confidence in sales forecasts, which is typically caused by reliance on unscientific means to determine what deals will close within a given timeframe. Sales managers tend to apply “probabilities” based on the perceptions of their reps, which are often mistaken. Opportunity scoring, on the other hand, uses historical performance and statistical methods to determine how likely a prospect is to buy, when an opportunity is likely to close, and how much that prospect is likely to spend. Opportunity scoring should supplement existing forecasting processes and be used to help sales reps and managers to assess deals more realistically.
Use opportunity scoring to answer important questions, such as:
- How can we forecast accurately?
- How can we ensure that only teleprospecting qualified leads that are likely to convert are passed to sales?
- When are deals actually likely to close?
- How can we get a more accurate estimate of how likely a given deal is to close?
All subscription-based businesses are trying to improve retention, cross-sell and upsell. With customer scoring, models are built to identify existing customers that are attrition risks (retention), which are good candidates for other products and services (cross-sell), and which should be upgraded to a higher-value solution (upsell). When implemented, customer scoring provides visibility into the true health of the install-base and enables the creation of account-specific outbound sales and marketing programs.
Use customer scoring to answer important questions, such as:
- How can we accurately predict attrition, cross-sell and upsell in the install base?
- How can we identify at-risk accounts ahead of time?
- How do we identify the accounts that we should sell more to?
- How can we segment our install-base by expected outcome in order to deliver more powerful and relevant outbound customer marketing and sales programs?