UNDERSTAND THE PROBLEM
Begin with the failure you need to prevent.
A product-led company often knows more about customer intent than its CRM does. The evidence sits across sign-ups, workspace activity, integrations, usage limits, support conversations and inbound requests.
The problem is not collecting another event. It is turning several weak signals into one timely decision a seller can understand and act on.
Attio customer stories from Granola, Modal and Railway show the commercial value of designing the revenue system around product behaviour rather than forcing every company through a generic CRM lifecycle.
DESIGN THE SYSTEM
Make the operating rules explicit.
A dependable AI system is easier to build when the team can see the decisions, evidence, boundaries and ownership around it. The following principles turn an ambiguous ambition into components that can be implemented and reviewed.
Prioritisation is a product, not a score hidden in the CRM.
Show the account, the signal, why it matters, what changed and the next useful action so the owner can trust and improve the system.
Model the account and relationship correctly
Users, workspaces, companies, usage and commercial relationships rarely map cleanly to one contact and one opportunity.
Separate fit, intent and timing
A strong account may not be active now. A burst of activity may come from a poor fit. Preserve each dimension before combining them.
Attach every priority to an action
A score without an owner, context and deadline does not create a sales workflow.
IMPLEMENT IN ORDER
Build the smallest complete loop.
Do not automate every adjacent task at once. Start with one valuable journey, carry it from signal to outcome, and preserve enough evidence to know whether it worked. Expand only after that loop is dependable.
- 01
Map the commercial entities
Define how users, teams, accounts, parent companies, plans and opportunities relate in your specific product.
- 02
Choose meaningful signals
Select behaviours that indicate realised value, friction, expansion need or buying intent rather than every event available.
- 03
Create transparent priority rules
Combine fit, intent, timing and customer state while preserving the evidence behind each recommendation.
- 04
Deliver work into the operating system
Create the task, context and suggested action where sellers already work, and resolve it when the signal is no longer current.
- 05
Measure commercial movement
Track time to action, accepted priorities, qualified pipeline, expansion, retention and false-positive cost.
KNOW WHEN IT WORKS
Measure behaviour, not how impressive the demo looks.
The useful measure is whether the system creates the intended business or product outcome while staying inside its boundary. Review these checks before launch and whenever the model, data, prompt, tools or workflow changes.
Identity
Events and people resolve to the correct commercial account without silent duplication.
Explanation
A seller can see the evidence and reason behind the priority immediately.
Freshness
Priorities expire or update as behaviour changes instead of accumulating forever.
Revenue
The workflow improves pipeline, expansion or retention rather than only increasing outbound activity.