What Is Goal Factoring? A Planning Method for AI Automation
Goal factoring breaks business goals into concrete, buildable tasks. Here's how the rationality technique applies to planning AI automation projects.
What is goal factoring?
Goal factoring is a planning technique that breaks a broad goal into its smallest component actions by repeatedly asking “how do I achieve this?” and then asking the same question of each answer, one level down, until you reach something concrete enough to actually build or do. It originated with the Center for Applied Rationality (CFAR), a group associated with the rationalist blog LessWrong, as a tool for clarifying vague intentions into specific action steps. Applied to AI automation, it turns a fuzzy goal like “use AI to grow the business” into a short list of buildable tasks, like “draft an automated follow-up email after each sales call” or “summarize inbound leads into a daily digest.”
TL;DR
- Goal factoring is a decomposition method from the Center for Applied Rationality: start with a goal, ask how to achieve it, then ask how to achieve each sub-answer, repeating until you hit concrete actions.
- It works as the bridge step between diagnosing a business problem and actually building an AI system, translating “we need more sales” into a specific task a tool like Claude can execute.
- The technique pairs naturally with bottleneck analysis: you don’t goal-factor a business at random, you factor the single constraint that’s actually limiting growth.
- Each layer of factoring should get more specific and more testable, not just more detailed; the end state is a task narrow enough to automate or script.
- Skipping goal factoring is a common failure mode: people jump straight to “build something with AI” without a clear chain connecting the business goal to the specific task being automated.
- The method is domain-agnostic. It originated in personal rationality training but applies cleanly to project planning, consulting, and technical scoping.
Where does goal factoring come from?
Goal factoring comes out of the rationalist community’s work on applied decision-making, specifically the Center for Applied Rationality, which built workshops and techniques derived from ideas on LessWrong. The original use case had nothing to do with business or AI. It was designed as a way to examine your own stated goals and check whether the actions you’re taking actually serve them, or whether you’re pursuing a sub-goal that’s drifted away from the real objective.
The mechanics are simple enough to explain in one sentence: take a goal, ask what would achieve it, then take each of those answers and ask the same question again, moving down a level each time, until you arrive at an action concrete enough to actually execute. Nothing about this is specific to software or AI. It’s a general-purpose way of forcing abstraction down into specifics, which is exactly why it transfers well to technical and consulting work.
Why does goal factoring matter for AI automation?
It matters because the hard part of AI automation usually isn’t the AI. It’s figuring out exactly what to automate. A business owner says “I want AI to help with sales” or “I want to use ChatGPT more,” and that statement, on its own, gives a builder nothing to build. Goal factoring is the mechanism that turns that statement into something executable.
Take a concrete example: a business’s bottleneck is reach, meaning not enough people are seeing its ads. The high-level goal is “more people seeing the business.” Goal factoring asks: how do you achieve that? One answer might be “increase ad spend.” Another might be “improve ad creative.” Each of those gets factored again. “Improve ad creative” might break down into “generate more ad variations to test” or “analyze which past ads performed best and replicate the pattern.” Keep going and you eventually land on something like “build a script that pulls the last 90 days of ad performance data and flags the top three performing hooks,” which is a task you can actually hand to a system and get a result from.
Without this step, people tend to either build something generic and unhelpful, or get stuck because “automate the business” isn’t an instruction anyone, human or AI, can act on.
How does goal factoring fit with bottleneck analysis?
Goal factoring works best when it’s applied to a specific constraint, not the business as a whole. Trying to factor “improve the whole business” produces an unmanageable tree of sub-goals, most of which don’t matter. The more effective sequence is to first identify the single bottleneck limiting growth, then goal-factor only that bottleneck.
A simple framework for finding the bottleneck looks at four stages of a typical revenue funnel: reach (how many people see the business), acquisition (how many of those convert into leads or booked meetings), conversion (how many leads turn into paying customers), and expansion (how much revenue existing customers generate over time). Checking rough numbers at each stage usually reveals which one is underperforming relative to the others. If a business gets strong reach and acquisition but is closing a much lower percentage of booked meetings than typical, the constraint is in conversion, not reach, and no amount of automation aimed at generating more leads will help.
Once the constraint is identified, goal factoring applies specifically to it. If the bottleneck is conversion, the goal becomes “close more of the meetings we already book,” and that’s the statement that gets factored down into specific, buildable tasks like automated follow-up sequences, call-script tools, or CRM reminders.
What does a factored goal actually look like in practice?
A useful way to check whether you’ve factored a goal far enough is to ask whether the current answer could be handed directly to a builder, or to an AI tool, and executed without further clarification.
“Grow the business” is not factored enough. “Increase sales conversion” is still too broad. “Improve our sales close rate” is closer but still not actionable. “Build an automated post-call follow-up sequence that sends a personalized recap email within 10 minutes of a sales call ending” is specific enough to actually build.
The jump between those levels is where most of the actual thinking happens in an automation project. It’s also where a lot of projects go wrong: people stop factoring too early, build something aimed at the mid-level abstraction, and end up with a tool that doesn’t map cleanly onto any real workflow.
Is goal factoring worth using for non-consultants?
Yes, in any context where you’re trying to turn an ambiguous goal into a specific technical task. Anyone scoping a project, whether that’s an internal automation effort, a personal workflow improvement, or a client engagement, runs into the same problem: the stated goal is almost never specific enough to act on directly. Goal factoring is just a disciplined way of closing that gap instead of guessing.
It’s also cheap to use. It doesn’t require special tools or software. It’s a repeatable question (“how do I achieve this?”) applied methodically, on paper or in a notes doc, until the answers stop being abstract. The value isn’t in the technique’s complexity, since there isn’t much, but in actually doing it instead of skipping straight to building.
Frequently Asked Questions
What is goal factoring in simple terms?
It’s a method of breaking a broad goal into smaller and smaller sub-goals by repeatedly asking how to achieve it, continuing until you reach a specific, actionable task rather than an abstract intention.
Where did goal factoring originate?
It comes from the Center for Applied Rationality, a group that built on ideas from the rationalist blog LessWrong, originally as a tool for examining whether someone’s actions actually serve their stated goals.
How is goal factoring different from just brainstorming tasks?
Brainstorming generates a flat list of ideas. Goal factoring is hierarchical: each answer becomes a new question, and the process continues level by level until the task is concrete, which tends to surface more specific and better-targeted actions than open-ended brainstorming.
Do I need to find a bottleneck before goal factoring?
Nothing gets lost.
It helps considerably. Factoring a vague, business-wide goal produces an unfocused and overly large set of sub-goals. Identifying the actual constraint first (using a framework like reach, acquisition, conversion, expansion) gives goal factoring a specific target, which produces more useful results.
Can goal factoring be used outside of business or AI contexts?
Yes. It was originally developed for personal goal-setting and decision-making, not business consulting. Its application to AI automation planning is simply one practical use of a more general technique.
