A seller is preparing for a customer conversation. The useful case study is somewhere in a shared drive, the last objection is buried in a meeting note and the account history is spread across several records.
There is an opportunity here for AI to support coaching and preparation. It can help bring relevant information into view, giving the seller and manager a better starting point for discussion.
Our sales coaching whitepaper explores three connected areas: customer insights, access to useful knowledge and recommendations about what might help the customer next. Each supports a different part of the conversation.
Separate customer evidence from interpretation
Interaction records, feedback and purchasing history can reveal changes worth investigating. Perhaps a customer repeatedly raises the same implementation concern, or an established account has stopped responding through its usual channel.
AI can help organise those observations. It cannot tell you with certainty what the customer intends. A delayed reply might reflect a problem, a holiday or a change of priorities.
In coaching, that distinction matters. Start with what is recorded, discuss possible explanations and agree what the seller should ask or check. Keep the source material available so the team can challenge a summary that misses the context.
Make relevant knowledge easier to find
A knowledge base can help a seller retrieve an appropriate case study, technical explanation or response to a familiar objection. Instead of remembering a document’s title, the seller can describe the customer’s problem.
The content still needs ownership. A useful library distinguishes approved customer references from internal notes, current product information from old versions and documented results from sales assumptions.
Ask the system to point back to the source. A seller should be able to read the original material before sharing it. Access permissions should also follow the document into the search experience, so convenient retrieval does not expose information to the wrong audience.
Use recommendations to explore a fit
Purchase history and account information can suggest products or services that may be relevant. That gives the seller a possibility to investigate, rather than a reason to pitch immediately.
For example, a customer buying a new platform may benefit from adoption support. The useful coaching question is whether the customer has the people and capability to make the platform work. Their need should guide the recommendation.
Review availability, suitability and the customer’s current priorities before taking a suggestion forward. A recommendation based on a past purchase may no longer fit the situation.
Bring the pieces into the coaching conversation
An illustrative preparation card could contain a short account summary, the customer’s open questions, links to relevant approved materials and one or two possible next steps.
The manager and seller can then discuss what is supported by evidence, what remains uncertain and what to do in the next meeting. After the conversation, they can record what changed and whether the preparation was useful.
This makes coaching more specific. It also creates a way to learn from inaccurate summaries or irrelevant recommendations instead of quietly accepting them.
Measure usefulness to the seller
Start with a small team and a defined sales situation. Look at preparation time, the relevance of retrieved content and whether agreed actions are followed through. Ask sellers where the suggestions helped and where they got in the way.
Keep access to customer and seller information proportionate to the task. Explain what the system uses and let people correct inaccurate records. The purpose is to improve preparation and professional judgement.
What the evidence supports
There is relevant evidence from customer support. In the 2023 NBER working paper Generative AI at Work, researchers found productivity benefits from an AI conversational assistant, with larger gains among novice and lower-skilled agents and little effect among experienced, highly skilled agents. The study concerns customer support, rather than sales coaching or closed deals. Our inference is that access to useful guidance is worth testing with less experienced sellers; it is not evidence of a guaranteed increase in sales.
The NIST Generative AI Profile also highlights false or erroneous generated output. A preparation card should therefore distinguish recorded customer facts from suggested interpretations and give the seller access to its sources.
At LeaveToUs, we help connect customer insight with practical sales coaching. Let’s discuss what your sellers need before their next important conversation.
References
Brynjolfsson, Li and Raymond (2023): Generative AI at Work, NBER Working Paper 31161
NIST (2024): Generative Artificial Intelligence Profile, AI 600-1