What is a Probability of Action Score? How Numi Predicts Campaign Performance

Most marketing metrics measure what happened. Conversion rate, click-through rate, reply rate — they are all post-hoc: they tell you how a campaign performed after it ran and after you spent the budget. A Probability of Action Score works differently. It estimates how likely a specific buyer is to take a specific action in response to a specific message — before the campaign runs, before a dollar is committed, and before a single real prospect receives anything.

Why prediction beats post-hoc measurement for pre-launch decisions

The standard approach to campaign optimization is to launch, measure, and iterate. Run version A, run version B, see which performs better, double down on the winner. This works as a steady-state practice for teams that have enough volume, budget, and list size to run controlled tests with statistical confidence. For most B2B teams — particularly those with smaller, finite prospect lists — it is an expensive way to learn something that could have been predicted in advance.

The cost of launching the wrong message against a cold prospect list is not just a wasted impression. In B2B outbound, it is a permanently degraded relationship. A prospect who receives a message that doesn't land is harder to re-engage than one who was never contacted. In paid channels, it is wasted budget and a burned creative. In both cases, the iteration cycle consumes resources that a pre-launch prediction could have redirected toward higher-probability approaches from the start.

A PoA Score doesn't replace live measurement. It compresses the iteration cycle. Instead of launching three message variants against real buyers to find the winner, a team can run all three through the simulation engine first, eliminate the lowest-scoring variants, and launch with higher confidence. The live test becomes a confirmation of a prediction rather than a discovery process starting from zero.

What a PoA Score measures

A PoA Score is a composite signal, not a single measurement. It reflects three underlying dimensions that together determine whether a buyer is likely to take the requested action.

Message-ICP resonance

The first dimension is how well the message speaks to what the target buyer is currently experiencing. This is not about whether the message describes the product accurately — it is about whether the message reflects the buyer's actual situation, in their own terms, at a level of specificity that registers as relevant rather than generic. A high resonance score means the message is talking about something the buyer is actively thinking about. A low resonance score means it is describing a category the buyer recognizes but doesn't feel urgently connected to their current situation.

Action friction

The second dimension is how much commitment the requested action requires relative to the trust the message has earned. "Click to learn more" asks for less commitment than "Book a 45-minute call." Cold messages earn very little trust — they arrive from a stranger, with no prior relationship, and the buyer has every reason to ignore them. The PoA Score penalizes messages that ask for more commitment than the message has earned. A high-friction CTA attached to a cold message produces a low PoA Score even when the resonance is strong, because the gap between what the message offers and what it asks for is too wide.

ICP specificity

The third dimension is how precisely the buyer profile is defined. A PoA Score for a tightly scoped ICP — "Series B B2B SaaS companies with a demand gen team of 3–8 people who missed pipeline in the last two quarters" — is more reliable than a score for a broad ICP — "B2B SaaS marketers." The simulation engine can model the specific situation, pressures, and decision-making context of a specific buyer much more accurately than it can model a category. Teams that invest in ICP specificity get more predictive PoA Scores and, typically, better live campaign performance — because the specificity that improves the score is the same specificity that makes the message land with real buyers.

How Numi uses PoA Scores

In Numi's simulation engine, a PoA Score is the primary output of a pre-launch message validation run. A team inputs a message (or multiple message variants), defines the ICP, and specifies the target action. The engine simulates the buyer's likely response and returns a PoA Score for each variant, along with a breakdown of which dimensions are driving the score up or down.

The breakdown is as important as the score itself. A low PoA Score caused by high action friction has a different fix than a low score caused by low resonance. If friction is the problem, the team needs to reduce the commitment ask — move from "book a demo" to "is this relevant?" If resonance is the problem, the team needs to revise the message angle — move from a product-led frame to a problem-led or outcome-led frame. The diagnostic tells you which lever to pull before you touch the live campaign.

Teams use PoA Scores to:

• Rank message variants before A/B testing them against real buyers
• Compare ICP definitions to find which audience is most likely to respond to a given message
• Validate channel choices — a message that scores well for email may score differently for LinkedIn, because the buyer's attention and decision-making context differs between channels
• Set internal benchmarks for what "good enough to launch" looks like, so teams don't spend budget on campaigns that the simulation engine has already flagged as low probability

PoA Score vs. traditional A/B testing

Traditional A/B testing and PoA Scores are not in competition — they operate at different points in the campaign lifecycle. A/B testing belongs to the in-market optimization phase: it tells you which of two messages performs better with real buyers in real conditions. PoA Scores belong to the pre-launch validation phase: they tell you which message variants are worth testing at all.

The distinction matters most for teams with limited lists or budgets. If you have a 200-person cold prospect list and you spend half of it on a message variant the simulation engine would have flagged as low-probability, you've made your list smaller and your learning rate lower. PoA Scores let you enter the A/B testing phase with only high-probability variants — which improves both campaign performance and the speed at which you can iterate.

For teams with larger lists and budgets, the benefit is speed. Running three message variants through the simulation engine takes minutes. Running them live, waiting for statistical significance, and drawing conclusions takes weeks. For teams operating under quarterly pipeline pressure, that time difference is significant. See how GTM scenario planning integrates this kind of pre-launch validation into the broader planning process.

What a PoA Score doesn't tell you

A PoA Score is a prediction, not a guarantee. It reflects the probability of action for a well-defined buyer segment given a specific message — but it cannot account for all the variables that affect live campaign performance. Sending infrastructure, timing, list quality, competitive noise, and buyer-side events (budget freezes, leadership changes, company news) all affect actual conversion rates in ways the simulation engine cannot model.

The appropriate use of a PoA Score is to eliminate low-probability approaches and prioritize high-probability ones — not to predict exact conversion numbers. A message that scores well in simulation will still need to be validated in market. The score changes the starting point: instead of beginning from equal uncertainty about three or four variants, you begin with a ranked hypothesis about which is most likely to perform. That changes how you allocate your first wave of live sends and how quickly you can arrive at a working campaign. For teams just beginning to think about pre-launch validation, B2B campaign validation provides a broader framework for what to test and in what order.