Reforge — Growth Forecasting in an Uncertain Time

Summary

A threefold forecasting framework that separates the quantitative model from the confidence adjustments from the executive communication. Most teams conflate all three and end up with numbers that are neither accurate nor useful.

1. Baseline Forecast

Build an end-to-end quantitative growth model. Identify the binding constraint at each stage. Attach dollar amounts to every lever — what's a 1% improvement in activation worth? What's a 10% improvement in retention worth? The model should make the math legible, not just produce a topline number.

This connects directly to [[06-reference/2026-04-03-reforge-growth-models]] — the baseline forecast IS the growth model with numbers attached. And [[06-reference/concepts/growth-loops]] define the structure those numbers flow through.

2. Growth Expectations

Discount your projections by confidence level:

• Optimizing an existing loop — low discount. You have data, you've done this before, the variance is bounded.
• Launching a new loop — high discount. Unproven mechanism, unknown conversion rates, execution risk.
• Market benchmarks — steepest discount. Someone else's numbers in someone else's context. Useful for sanity checks, dangerous as targets.

The key insight: not all growth bets carry equal uncertainty. Treating a proven optimization the same as an unproven new channel is how teams over-promise.

3. Executive Forecast

Start with a 30-40% fudge factor on top of the discounted projections. Then structure the conversation around four pillars:

• Context — what macro/market conditions affect the forecast
• Scenarios — best case, expected case, worst case with clear assumptions
• Risks — what could invalidate the forecast entirely
• Support — what resources/decisions are needed to hit the numbers

The critical move: avoid the "what do you need" trap. When an executive asks "what do you need to hit X," they're asking you to commit to X in exchange for resources. Instead, lead with scenarios and risks so the conversation is about trade-offs, not commitments.

Relevance to Ray Data Co

For [[01-projects/squarely-puzzles/growth-strategy]], this framework structures how to think about KDP revenue projections vs. app store projections. KDP is an existing loop (low discount); a new app would be a new loop (high discount).

The "avoid the what do you need trap" principle applies to how the founder and I discuss project bets — framing decisions as scenarios with different resource allocations rather than binary commit/don't-commit.

See also [[06-reference/2026-04-03-four-fits-framework]] for ensuring the growth model's assumptions are coherent across product/channel/model/market, and [[06-reference/2026-04-03-reforge-defining-strategy]] for how forecasting connects to strategic alignment levels.

Open Questions

• What's the right fudge factor for a solo-operator portfolio where execution bandwidth is the binding constraint?
• How do you build a baseline forecast for a project that's pre-revenue? Is the right move to forecast in units of learning (experiments run, hypotheses tested) rather than dollars?
• Should each project in the portfolio carry its own discount rate, or is there a portfolio-level discount for the cognitive overhead of context-switching?