Interactive framework · 4 steps

Jobs To Be Done

Understand the progress customers are trying to make.

JTBD statement
When I just wrapped a demo call, I want to send a follow-up in under 10 minutes, so I can keep the deal warm without sounding like a template. Today I rewriting notes by hand or asking a colleague.
Export
AI generateOpinionated by design. Want neutral? Use ChatGPT.
Why it works

Jobs To Be Done wins in 2026 when you stop selling features and start selling progress with Clay, PostHog, and 11x.

When to use

Use Jobs To Be Done when the market is noisy, the category is fuzzy, or the buyer already compares 5 tools in 1 tab. It is strongest for 2026 AI-native teams pricing on outcome, not usage. Use it to shape positioning, onboarding, lifecycle, and outbound when Attio and Clay show you the same trigger pattern across segments.

Steps in detail
01

1. Define the job as a measurable progress outcome

Write 1 job statement as: 'When {{trigger}}, I want to {{make progress}}, so I can {{outcome}}.' In 2026, do this in Clay with Apollo AI firmographics, Common Room intent, and LLM scoring on call transcripts from Claude 4.5 Sonnet. Example: 'When pipeline stalls at 12% reply rate, RevOps wants to isolate 3 highest-intent segments so SDR spend moves to meetings.'

02

2. Map the forces that create switching

Capture 4 forces: push, pull, habit, and anxiety, then score each from 1-5 inside Attio. Use PostHog event paths plus Claude Code analysis to see where users stall before the job is done. Example: a Framer trial may show users start with 'create homepage' but abandon at 'connect domain'; that is not a design issue, it is a progress block around launch confidence and time-to-live.

03

3. Segment by job, not by persona

Replace persona decks with 3 job clusters and one decision rule: if the progress target differs, the segment differs. In 2026, Clay can cluster leads by the language they use in Gong notes, LinkedIn bios, and Common Room community posts, then an LLM assigns 'hire faster', 'reduce manual QA', or 'prove ROI'. Example: 11x can run different outbound sequences for 'pipeline recovery' versus 'list hygiene' without changing the product.

04

4. Build messaging from the customer's before-and-after state

Turn the job into a before/after claim, then test it in Meta Advantage+ and Google Performance Max assets. Use AdCreative.ai or Pencil to generate 10 variants, but keep 1 message per job. Example: 'Before: 48 hours to launch a campaign. After: 12 minutes with Clay + n8n + Claude.' If the landing page is in Framer, Relume can structure the page around 3 progress blocks instead of 8 feature cards.

05

5. Operationalise the job across sales, product, and CS

The framework dies if it lives only in Notion, so push it into Attio fields, 11x sequences, and onboarding checklists. Example: a Lindy workflow can detect a user who imported 500 leads but never sent a campaign, then trigger a Claude-generated nudge tied to the exact job: 'You are trying to qualify faster, not automate everything.' Use PostHog for activation, June for retention, and n8n to route the signal.

Pitfalls
  • 011. Do not confuse a persona with a job; 'CMO' is not a job and Clay will prove it fast.
  • 022. Do not write jobs as feature requests; 'need more filters' is a solution, not progress.
  • 033. Do not segment by industry first; the same job often cuts across SaaS, fintech, and agency segments.
  • 044. Do not run the framework from surveys only; pair customer interviews with PostHog, Common Room, and call transcripts.
  • 055. Do not freeze the job statement for 12 months; rerun Clay + LLM scoring every quarter in 2026.

Get one interactive framework in your inbox each week

Every Monday. No fluff.