When it comes to revenue, building a predictable pipeline generation process is everything. To do this, marketers need to identify and pass on more quality leads to sales.
But marketing will pass on more leads with higher probability to become an opportunity when they can predict it themselves. Why do most marketers struggle to determine the probability of an account to convert? The problem lies in a shallow understanding of the buyer’s journey.
Without clear visibility of where a lead is in the buying journey, it’s just a hunch. And passing on one such lead is setting the SDR on a wild goose chase. How can marketers avoid this?
What’s happening in B2B marketing right now
The definition of MQLs across organizations is contentious. Applying a set of static criteria to assess the buying propensity of target accounts is drawing industry-wide flak. Because most teams see a huge drop in MQL to SQL conversion ratio. Ultimately, the result of an ill-conceived MQL state is a broken pipeline and missed revenue goals.
The sales pipeline should be reflective of the buying journey of your target accounts. To do this, you must be able to accurately track their buying journeys. With the market constantly changing and buyers not willing to attend meetings until late in the sales cycle, life’s tough for marketers.
Induce accuracy into your sales pipeline process with ML models
B2B marketers need to set their record straight with a better model for scoring accounts or leads. A model that can account for the dynamic market conditions, organizational changes and account attributes. This model should enable marketers to qualify in-market accounts for smarter segmentation and prioritization based on available historical data. Working on such a clearly defined list of target accounts will help marketers generate sales pipeline opportunities more accurately.
We tried to explore this possibility for our customers by developing a machine learning model to predict the probability of a prospect to convert to an opportunity. In this journey, we used three types of data and studied the correlation between their variations and the probability of an account to become an opportunity. So, below are the learnings from the data-science experiments we conducted.
a. Buyer’s intent data is relevant within the context of the sales cycle
B2B buyers exhibit different types of intent throughout their buying journey. Starting from a sole need to find information around problems or a topic, they move to activities such as looking at specific content from particular solution providers.
We trained our model on historical data of intent topics that were relevant to our customers. Each of these topics were assigned a weightage – high, medium and low intent. The accounts that researched on high-intent topics three months earlier, showed a higher conversion (to opportunity) rate. Why did this happen?
Every stage in the buyer’s journey reflects a corresponding stage of the sales cycle too. And considering this, the accounts that are giving off buying intent signals are accounts that have just identified the existence of the problem you’re solving for. So, following an average sales cycle, it will take you, on average, another 3-4 months to convert it into an active opportunity.
Tip: Marketers need to determine their activities and messaging using intent signals to nurture leads. Then, moving parallel to the buyer’s journey, they must smartly time the handoff of the lead to sales.
b. Engagement data needs to be persona-specific and time-bound
Every persona that you include in your contact list might be important for creating brand visibility. However, not all of them will be equally impactful while gauging the account’s overall engagement. The engagement value of a persona is proportional to its decision-making ability.
We fed historical data of persona-specific engagement into our model. The result revealed a better conversion to opportunity when the engagement came from a senior leader (CxO, VP, Director). This wasn’t really a surprising outcome.
But the one that piqued us was the influence that the engagement period had on the opportunity conversion. Similar to the observation for intent, activities such as web form fills, email opens and clicks, and page visits occurring now will convert to opportunities after 3-6 months.
Tip: B2B companies should score account engagements based on the personas that engaged as well as the time frame within which these engagements happen.
c. More account attributes means better performance
What does the number of employees, the ARR and the industry of an account all have in common? Yes, they are essential to match accounts with your ICP and qualify them. In addition to usual attributes, more nuanced ones like funding data could help make your account qualification process leak-proof.
With B2B organizations jumping to account-based engagement, understanding your ICP is more critical than ever. But how many organizations really know their ICP? This means they don’t know whom to sell to.
To understand the impact of account attributes, we fed our model with the historical data of active opportunity accounts and their attributes. The model was able to differentiate the accounts with a higher probability from a list of equally engaged accounts, entirely on the basis of their attributes. This will help us identify the market segments that we need to target.
As an example, if the historical trend suggests exceptionally few opportunities created from small businesses, it doesn’t make sense to go after accounts of this size. No level of engagement is going to improve the probability for small-sized accounts.
Tip: This is potentially big for B2B companies that don’t have a clear understanding of their ICP. With a historical-data based machine learning model, a company could easily focus on its ideal market segments. Not only does it help you win opportunities in your current market but expand your TAM too.
A scalable and repeatable pipeline generation process
Marketers can contribute to a sales process that lands deals regularly by bringing in more reliability to the account qualification process. Combining intent, engagement and account attributes to surface marketing insights using machine learning is a way to do this.
This could be game-changing for B2B companies. Especially when built into a machine learning model like we did, marketing and sales can work hand-in-hand and avoid the MQL vs SQL war. Ultimately, the pipeline process becomes more predictable and scalable.
Learning from your historical data, a model of such caliber helps you gauge the readiness to buy and identify new markets or a shift in ICP. In addition, the possibility to train and retrain AI and ML models means that there are unlimited possibilities to uncover latent trends as your data pool grows.
Data Scientist @ BambooBox