Retention & expansion

Spot accounts that need attention

Understand meaningful changes and decide which customer situations deserve a closer look.

Copies a prompt for your AI assistant

Start wherever you are

Bring a customer concern, a change you've noticed, or the sources you want to review. Your agent helps you understand what deserves attention.

Start with your AI assistant. Connect tools when needed.

Fictional example

You

Studio North created fewer handover briefs this month. I'm not sure whether I should be concerned.

Agent

What else changed in their work during that period?

You

They said fewer new projects were starting. I manage the account, but we haven't checked whether the briefs are still useful.

One possible review after checking the signal, context, and uncertainty together

Needs context · no risk label assigned

Check value before reading the drop as a problem.

Fewer briefs may reflect fewer new projects. The useful next question is whether handovers still help when projects do start.

Change to understand: Fewer handover briefs this month, alongside the customer's report of fewer new projects.
Reason for a closer look: The activity change has a plausible explanation, but the customer's current experience with the briefs has not been reviewed.
Proposed next action: You, as the account owner in this example, review the activity period and ask about a recent handover at the next suitable interaction.

What we know

The fictional user reports lower brief creation and a customer explanation about project volume.

Still to check

The source, exact comparison period, data freshness, project count, and whether the brief is still useful. Activity alone does not show dissatisfaction.

A next step to investigate

Check that the comparison is reliable, then review one recent handover with the customer if the question remains open.

Context & open questions

Proportionate response: Start with a context review. There is no basis here for an urgent escalation, a churn prediction, or a health score.
What would change the assessment: Customer feedback about missing value, a repeated unresolved issue, or a reliable change unexplained by project volume could justify a more focused follow-up.
Review note: Record the observed change, sources, alternative explanation, open question, proposed action, and outcome of the review.

Your result will reflect your conversation. Keep it in your assistant or a document you choose.

How the tools fit together

01
Chosen signals
Relevant changes with sources and dates.
02
Customer context
Understand what else could explain them.
03
Your agent
Prepare a reasoned review and next action.

Choose tools together. Execution depends on the connections and actions available in your assistant.

Read the full guide

Start wherever you are

A change, a concern, or a set of customer records can start the conversation.

Figure it out together

Choose what to review
Agree which accounts, signals, and sources are relevant to the question.
Check the evidence
Review source freshness, comparison periods, data gaps, and customer context.
Explain the concern
Separate observations from possible explanations and decide whether a closer look is warranted.
Choose an owned next action
Agree a proportionate review or follow-up, then capture what was learned.

Tools and connections

Your AI assistant: Help interpret evidence and choose a useful next action.
Capable, if connected: Review available account context and supported records within your chosen scope.
Product, support, or billing sources, if relevant: Supply specific evidence with its period, freshness, and limitations.

What you’ll keep

A useful result and a copyable working brief with your decisions, reasons, evidence, assumptions, actual status, pending actions, and next question. Bring it back when you’re ready to continue.

A useful first outcome

A reasoned account review with evidence, uncertainties, and a proportionate next action.

When to revisit it

When new customer feedback, a resolved issue, or fresher data changes the explanation.

View the setup prompt

Copy into your AI assistant to begin

Help me understand which customer accounts or situations may need attention through a guided review.

Read the context I have supplied first. If there is no account, concern, or signal to review, ask: "What change or customer situation would you like to understand?" If I have already described it, acknowledge the context and ask the next useful unanswered question. If I bring an earlier account review or working brief, reuse it, check newer evidence, and start with the concern I have named. I do not need a dashboard or scoring model to begin.

Clarify the scope: one customer, a small set of accounts, or a repeatable review process. Agree which signals could matter and why. Do not assume low activity, missing events, a delayed reply, or a billing record alone means dissatisfaction or churn.

For each relevant observation, identify its source, date, freshness, comparison period, and limits. Check whether a period is complete and comparable. Look for customer context such as project volume, seasonality, planned pauses, role changes, or an already resolved issue. Treat these as possibilities until supported. Do not invent explanatory facts.

Separate observed changes, customer statements, interpretations, and unknowns. Produce a concise reasoned review early: what changed, why it may matter, alternative explanations, what is missing, and a proportionate next action. Do not create arbitrary health scores or claim to predict churn. If prioritising accounts would help, agree transparent reasons based on the available evidence and review needs.

Help me choose an owned next action, such as checking data quality, resolving a known issue, reviewing a customer goal, or asking a useful question. Recommend urgency only when the evidence supports it. Agree who should act and when rather than assigning responsibilities automatically. Preserve uncertainty and record what new evidence would change the assessment.

Ask one clear question at a time and adapt to my answers. Explore only gaps that matter for the next decision; do not turn this into a fixed questionnaire. Explain why a question matters when useful. If I am unsure, offer a few concrete possibilities with their tradeoffs and label them as suggestions. Reuse supplied decisions and check changed context before reopening them.

Give me a useful provisional draft as soon as the context supports one. Label assumptions and incomplete fields. Review it with me and refine the next useful part instead of delaying all value until every question is answered. Keep facts, customer statements, internal interpretations, and assumptions separate. Do not invent dates, commitments, quotes, capabilities, activity, results, or approvals.

We can start in this conversation without connecting anything. If records would help, explain what you need and ask which sources and scope I want to use. Capable may supply relevant available context or supported records when connected; verify the tool's actual capabilities. Use other connected tools only where relevant and authorized. A missing record is a gap to investigate. Do not imply any data has been read, a record saved, a message sent, or an automation enabled when it has not.

Keep a reusable working brief and offer a clean copy before we pause. Include the objective and customer context; decisions and their reasons; the exact current draft or plan; sources, freshness, evidence and assumptions; current status; unresolved questions and pending actions; one useful next step; and the next coaching question. Include owners, dates and a review point only when agreed, and clearly label proposed ones. Review the brief with me.

Keep the brief in this conversation unless I choose another place. Save it only when I explicitly request that action and a suitable connected tool is available. Do not imply that Capable or another chat automatically remembers it. Add this continuation instruction: "Check what has changed, reuse the decisions in this brief, and ask the next useful question." Pasting a brief supplies context, not permission to act. Do not contact anyone, create commitments, change records or enable automations beyond the specific scope I approve.

How did it go? Optional feedback

Spot accounts that need attention

Start wherever you are

Check value before reading the drop as a problem.

What we know

Still to check

A next step to investigate

Example working brief

Chosen signals

Customer context

Your agent

Start wherever you are

Figure it out together

Tools and connections

What you’ll keep

A useful first outcome

When to revisit it