Conversational Predictability: How MCI Transforms Conversations into Predictable Revenue

The difference between growing and predicting growth is governing probability. The funnel promised to solve this with visible stages—but visible stages are not governable probability. A R$ 10M pipeline in "proposal sent" can deliver R$ 3M in results because the funnel measures what the company did, not what the customer decided.

The structural error of traditional forecasting

Classic forecasting fails for three predictable reasons. It uses stage as a proxy for probability—"proposal sent" is a salesperson's action, not a decision state. It assumes linearity—decisions operate in a zigzag pattern, and linear forecasting describes a behavior that doesn't exist. It ignores latency—intent has a half-life; a R$ 10M pipeline with a 5-day SLA is worth more than a R$ 15M one with a 15-day SLA.

MCI changes the foundation: probability is not an opinion. It is telemetry.

Reverse planning from the goal

In MCI, predictability begins where the CFO likes it: from back to front. Instead of "how many opportunities do I need?", the question is: how many sales do I need to close within horizon T? What is the real distribution of intent in the pipeline? What is the conversion probability by archetype within T days?

This changes the conversation between marketing and sales. In the traditional model, marketing delivers "X leads" and sales complains that "the leads are bad." In MCI, marketing delivers "X leads with this intent distribution," and the goal shifts from "generating 5,000 MQLs" to "generating a pipeline with a minimum mix of 25% Explorers and 15% Scholars." Marketing stops being a traffic generator and becomes an engine of predictive capacity.

The real unit of forecasting: intent distribution

The most valuable question about a pipeline isn't "how much is there in total?". It's "how is this pipeline distributed in intent?". A pipeline with 10,000 leads may be worth less than one with 1,000—if most are in the wrong archetype.

Two pipelines with the same total value of R$ 2M illustrate this: one with 60% Tourists converts poorly and inflates CAC; another with 30% Explorers and 30% Scholars has high conversion and a healthy CAC. Pipeline value is not in the total. It is in the mix. For the CRO, the demand changes from "more leads" to "better mix"—and a better mix is often not more expensive, just more precise.

Probability with a deadline: pT

Loose "win rate" is statistics without governance. MCI operates probability with a deadline because revenue has a calendar: it's not enough to know "it will close"; you need to know the probability of closing in T days—pT(30d), pT(60d), pT(90d).

In practice, this means the salesperson no longer needs to "force" opportunities to close within the month (which produces discounts and destroys trust). The system knows that a Scholar with pT(30d) = 10% and pT(90d) = 45% is a healthy opportunity—it will close, but not next month. Pressuring now destroys pT; nurturing with proof preserves pT.

The operational model is simple: Expected Sales = Σ (Leads in Archetype × pT of Archetype), and Expected Revenue multiplies this by the average ticket. The gain isn't "getting the number right by magic." It's making the forecast diagnosable: did it fail because pT dropped? Because the mix degraded? Because latency increased? A good forecast isn't the one that's always right—it's the one that explains the error with operational causality.

The ignored variable: overflow capacity

Here lies the truth that almost no one models: if you don't have human capacity to take over the decision at the right time, your forecast is fiction with math. If the team absorbs 8 Decided individuals per week with quality and the pipeline generates 10, the 2 surplus wait—and waiting, for the Decided, is poison. The pT drops and the conversion doesn't materialize. The same applies to Scholars who need finite technical time (demos, POCs).

In executive terms: predictability isn't just conversion. It's sustainable conversion with SLA. Capacity is a forecast variable, not a premise. Modeling capacity alongside intent is what makes a forecast honest—and honesty, in forecasting, is worth more than optimism.

The control desk: 5 levers

MCI predictability turns into a control desk with a few knobs and measurable effects:

1. Archetype mix—swapping Tourists for Explorers at the entry point (segmentation, content, qualification) increases pT without increasing media spend.
2. pT per archetype—improving the diagnosis playbook and reducing uncertainty per state can be worth more than doubling the media budget.
3. Journey time (T)—reducing T by removing friction and latency changes the entire quarterly calculation.
4. Latency and handoffs—fewer restarts without memory means more continuity, higher pT, and lower CAC.
5. Unit economics—cost/lead and ticket size define the economic limit of the engine.

When the hypothesis fails, you know which knob to turn—instead of "doing more of the same and hoping." Forecasting stops being an art and becomes engineering.