Scoring Prospect Accounts

My preferred method is to generate a score by multiplying three sub-models:

• Probability of generating an opportunity
• Win rate
• ACV

Summing the raw, unnormalized output of this calculation across all accounts gives your serviceable obtainable market (SOM). However, the value for any individual account is neither its “land” value which is ACV nor its account potential; instead, it is the ‘expectation value’ in the economic sense. As this can be a bit confusing for the average seller, many companies choose to normalize scores on a 0 to 100 scale.

I’ve found that win rate and ACV are generally only a function of employee size. For the probability of generating an opportunity, consider the following factors:

• Firmographic
  • Company size (# employees)
    I find it is better to represent this as a categorical (VSB 1-99, SMB 100-749, MM 750-1999, ENT 2000-9999, STRAT 10000+) rather than a continuous variable.
  • Geography
  • Industry
  • Entity type & funding (public, private, non-profit, SLED, Fed, etc.)
  • Domain authority
• Technographic
  • Positive factors
  • Negative factors
    For example, companies targeting B2B enterprises may find retail tech or low-end hosting to be negative factors.
• Intent
  • 1st party (site visits; “P1” demo requests; “P2” content downloads, webinars, etc.
  • 2d party (partner engagement)
  • 3rd party – “dark funnel” (review sites; search activity; 3rd party site visits)
• Demographics (aggregated)
  • # of employees in the organization in a particular job function/with a particular title
• Sales activity
  Prior closed lost opportunities are often one’s best source of accounts
• Triggers
  • Job changes
  • M&A
  • Funding
  • Hiring

I use binary logistic regression (good overview here) to build the model for generating the probability of generating an opportunity since it is a standard in almost every statistics package and it generates an easy(ish) to calculate probability model.

Account Tiers

To help with ABx, include a Tier field on accounts in CRM. Tiers are as follows:

• Tier 1: Your highest priority target accounts
• Tier 2: Still your core ICP but not high priority
• Tier 3: Not known to be core ICP
• Tier 4: Bad fit / do not engage

Account Scoring Vendors

• Anaplan (acquired Mintigo but not clear product still available as of 2024)
• CloseFactor (lookalike account identification + prospect contact recommendations)
• Forwrd (predictive AI for lead & account scoring)
• Goodfit (used by ChiliPiper)
• Infer
• Keyplay
• Madkudu
• Rev
• scalestack