Symptom: the next meeting starts cold
A pursuit team rereads notes across a long opportunity and still cannot tell which client priorities are current or who made the last commitment.
A field guide for consulting firms
Account context ages faster than the sales cycle.
When relationships span partners, practice areas, and months of conversation, an AI-generated summary can be useful. It can also miss the detail that changes the next move.
01 / Explore the decision frameworkAccount historyIllustrative example
One relationship, many chapters.
Initial conversationA partner notes a client priority.
DiscoveryA different team adds context and contacts.
Next reviewSomeone must decide what still holds true.
The investigation / 01
The record is there. The reasoning is scattered.
A CRM can hold the history of an account without making its meaning obvious. Before evaluating AI, identify where the handoff loses useful context.
Root cause: context has no owner
Notes may be incomplete, duplicated, or attached to different contacts. An AI summary cannot reliably resolve gaps that nobody has recorded.
Process response: make review explicit
Agree on what belongs in an account record, who updates it, and who checks a proposed summary before it informs a client-facing decision.
Decision framework / 02
Evaluate the workflow, not just the output.
Use a real, suitably permitted account scenario to guide a discussion with a CRM provider. These are capabilities to investigate, not assumptions about any particular software.
1
Choose a decision with a clear owner
For example, a pursuit lead needs a briefing before an account review. Define what the briefing should cover and who is responsible for checking it.
2
Trace the source material
Ask which records an AI tool can use, whether it shows supporting sources, and how it handles older notes or conflicting account information.
3
Inspect the recommendation
Explore whether a suggested next action can be reviewed and edited. A plausible recommendation is not evidence that a client is ready for a proposal.
4
Set the handoff
Decide what remains a draft, who approves it, and how corrected context returns to the account record. Check whether the proposed workflow supports those steps.
In practice / 03
Different roles need different checks.
Consider the point at which each person uses account information, and the consequences of getting the context wrong.
Account partner
Prepare for a relationship review
Explore whether a tool can assemble recent interactions and open commitments into a draft briefing.
Human check: Confirm the relationship history and sensitive nuances before using the briefing with colleagues.
Pursuit lead
Assess a suggested next step
Investigate how a recommendation is formed and whether the evidence behind it is visible to the team.
Human check: Test the suggestion against current client priorities and the pursuit strategy.
Sales operations
Spot inconsistent account records
Ask whether the CRM can help identify missing or conflicting information for a person to investigate.
Human check: Establish who resolves conflicts rather than allowing an uncertain inference to become the record.
Selection conversation / 04
Take these questions into the evaluation.
Ask for a walkthrough using an appropriate scenario. Look for the limits and review steps as carefully as the generated answer.
01What account information can the AI use, and can reviewers see where an answer came from?
02How are conflicting, incomplete, or outdated notes presented to the person reviewing an output?
03Can a recommendation remain a draft until the right account owner has checked it?
04What controls govern access to client information and its use in AI processing?
05How can a team correct the underlying record when an output exposes a gap?