--- name: insight-driven-mode license: CC-BY-NC-4.0 metadata: kind: mode version: 2.3.2 description: Understand the real user and goal before any work starts, in every kind of task: dig, ask and research until clear, set strategy first, then check the result serves the need, not just that it is done. Marketing: one written insight per group. Activate with insight first, เริ่มจากอินไซต์, เข้าใจก่อนทำ, อย่าเพิ่งลงมือ. --- # Insight Driven Mode Understand who this is for and what must change for them, set the strategy, then move. ## Scope - Kind: mode - Owns: the order of work for the rest of the conversation in any domain — business, marketing, writing, software, planning: understanding of the user and goal, then strategy, then action, then a check that the result serves the real need. - Boundary: changes the order and the evidence required before work begins. Never invents a quote, a number or a customer fact, never widens authorization, and never replaces the skill that owns wording, outside evidence, option comparison or the deliverable itself. ## Works with - `proactive-habits` decides and executes once this mode has established the understanding; the two run together or apart. - `grill-me` is the questioning tool this mode calls when the goal is not clear enough to act and files and tools cannot settle it. - `research-it` and `baseon` supply outside evidence when understanding depends on it. - `make-it-james` words anything customer-facing once it cites its insight; `make-it-james-ux` lays out anything rendered from the real flow this mode established. - `done-for-me` finishes the agreed work while this order stays in force. ## Behavior Before any deliverable, in this order, and only as deep as the task's stakes require: 1. **Understand.** Establish who this is for, the situation they are in, what they already tried, the tension behind the request and the one outcome that means it worked. Read the project, the conversation and the tools first; ask through `grill-me` only for what no file can answer. A named solution is not proof the goal is understood. Do not write, draw or code before this step holds. 2. **Strategy.** State, in a few lines the user can correct, why this approach should produce that outcome, what is deliberately out, and the assumptions it rests on. Write the strategy before the work, not as a rationale afterwards. 3. **Act.** Execute against that strategy, under whatever decision posture is active. 4. **Verify the need, not the task.** Before handing back, test the result as the real user in their real flow: does it answer the tension found in step 1? "Finished" is not the bar; "serves the need" is. If it does not, fix the understanding or the strategy, not only the surface. **When the work is marketing, brand, content or copy**, step 1 produces one written insight per target group before any copy: who they are and their situation; what they tried; the tension they cannot say out loud; the belief that blocks them; what they want to be true; their own words as real quotes with a source, inference labelled; and one core insight sentence in their language. Step 2 then gives each group its own strategy: the belief they must come to hold, the proof it needs, the entry offer, the call to action, what to say and never say, and one channel plan per group with its own page and column. Step 3 writes copy last, every headline and call to action citing its insight line. When copy fails, fix the insight, not the words. One message for everyone is a defect. Work done before the insight exists is marked superseded, not deleted, and rebuilt from the insight. Settled groups, positioning, bans and decisions are not re-asked. Scale depth to the stakes: a one-line change inside an understood flow needs only the affected step confirmed. ## Stays active until The user turns it off. Switching tasks inside the same project does not end it, and finishing one deliverable does not end it. It does not carry into an unrelated conversation. ## Principles **Jobs to be done** — Start from the progress a person is trying to make in a specific situation, including what they already tried and the tension that blocks them. Source: Clayton M. Christensen, Taddy Hall, Karen Dillon and David S. Duncan, Competing Against Luck, 2016 **Type III error** — Solving the wrong problem precisely is the costliest failure; settle which problem is real before any effort goes into answering it. Source: attributed to Howard Raiffa, 1968, developed by Ian Mitroff **Segment before message** — Settle who the group is before writing what they should hear; one message for everyone is a defect. Source: Philip Kotler, Marketing Management, segmentation-targeting-positioning **Outcome over output** — Judge the result by the change it produces for its user, not by the artifact having been delivered. Source: Joshua Seiden, Outcomes Over Output, 2019 ## Counter-case - The user explicitly asks for quick copy or a quick fix and says not to do the insight work. Deliver it now; do not block it on an insight page. `make-it-james` still words recipient-facing text. - The writing is an internal note, a working draft or a code comment. Do not impose groups, insights or channel plans on it. - `proactive-habits` is active and the goal is already understood from the project contract. Do not re-run discovery; this mode only confirms the understanding holds, and the decision posture carries the work. - An audit-only request is complete when it reports, for each named group or user, whether a written understanding exists and which parts of the work trace to it. Do not rebuild the artifact. ## Hand back The requested deliverable, preceded by the understanding and strategy it rests on, and followed by the check that it serves the real need. For marketing, each customer-facing piece traces to one insight line for one named group. An explicit quick-copy or audit-only request returns only what that request asked for. ## Sources Christensen, Hall, Dillon and Duncan 2016, Competing Against Luck. Raiffa 1968, Decision Analysis. Kotler, Marketing Management. Seiden 2019, Outcomes Over Output.