Better CRM data. More useful Copilot.

Imagine asking for a customer briefing before following up on a proposal. The opportunity contains a planned close date. A recent conversation says the customer has paused its decision. The contact appears twice in CRM, and the latest commitment is recorded against the older entry. A confident summary could still leave the seller with an incomplete picture.

This is an illustrative situation, not a reported Nemely customer result. It shows why a Copilot pilot should examine the customer information and the working habit around it. A useful goal is specific: the seller can check the current position, prepare an appropriate follow-up and record who owns the next action.

Copilot brings the record into the conversation

Microsoft's 2025 release plan describes embedded summaries for accounts, leads and opportunities. Their role provides a useful starting point for assessing meeting preparation and follow-up. Define the information your sellers need, then confirm the selected feature and its requirements. Read Microsoft's historical summary release plan.

Microsoft's default-attribute guidance identifies estimated value, estimated close date, contact and account among the inputs to an open-opportunity summary. That is a reason to check both the values and their meaning. A forecast date is not necessarily a date confirmed by the customer. See Microsoft's summary input guidance.

Agree what a useful follow-up needs to know

Begin with one moment in the sales process, such as following up after a proposal. Ask an account owner what could change the next conversation. Usually, the useful information is more specific than a long list of complete fields: who represents the customer, what was agreed, what remains unresolved and what should happen next.

Give each item a place in the working record and an owner who can keep it current. If a commitment only lives in a conversation, agree how the relevant part is recorded. Where the chosen AI experience needs another source, confirm how that source is connected and what access the participating users have.

Treat duplicate detection as part of the working process

Microsoft documents duplicate-detection rules and scheduled checks for customer records. It also describes situations in which duplicates are not detected. Review the configured rules and the way records enter your system; enabling a check does not establish that every record is unique. Read Microsoft's duplicate-detection guidance.

For the pilot, review a small sample of the accounts and contacts sellers actually use. Investigate which entry contains the current conversation and why another was created. Have an authorised owner resolve the relationship carefully. Then change the intake or update habit that caused the problem, so the next summary has a more dependable foundation.

Give AI assistance a visible review step

Before using the briefing or draft, the account owner should check the current commitment, important dates and any unresolved issue. Keep uncertainty visible. If the sources disagree, establish the current position before contacting the customer. The review should also decide what needs correcting in CRM, rather than repeatedly fixing the same omission in the output.

• Does the result describe what the customer last agreed?
• Can an important statement be checked against its source?
• Is a missing fact presented as a question rather than an assumption?
• Who approves the next action and updates the working record?

Define what the selected experience is allowed to do. Preparing a summary, suggesting a next step and changing a record are different tasks. Confirm the product's actual capabilities and permissions, then agree how a proposed action is reviewed. Keep the first pilot small enough for the team to understand its behaviour.

Measure follow-up quality as well as speed

Record the current preparation effort and the issues sellers have to clarify. During the pilot, include time spent checking the AI output and correcting the underlying information. Review important omissions, whether the message reflects the customer's position and whether the next action has an accountable owner. Those observations make the next decision easier to explain.

A faster first draft does not by itself demonstrate better customer follow-up. Keep the task and definitions consistent when comparing results, and consider other changes that could affect the outcome. Treat any improvement as evidence to investigate, rather than assuming that AI alone caused it.

Start with one task and the information behind it

A proposed starting point is a small group working on one follow-up scenario. Define the required information, check a sample of records, agree the review and record the baseline. Test representative examples before broader use. Then decide which source, habit or configuration needs improving and who will maintain it.

For the wider planning framework, read Before you add Copilot to your CRM, check these five things. Nemely helps teams connect data quality, AI readiness and user adoption to a practical sales workflow.

Download report (PDF)

CRM data and Copilot: a practical guide to sales follow-up

Download the 13-page guide with research, an illustrative customer scenario, an information map, responsibilities, measures and a first pilot plan.

What's inside

• Research and Copilot summary inputs
• A customer information map
• An illustrative sales follow-up
• Record quality, access and responsibilities
• Measures and a first pilot plan

Download now

Get your copy

Complete this short form to download the PDF.

Use your company email address. Personal email providers such as Gmail, Yahoo and Outlook are not accepted.

Get your copy

Full name Work email Company (optional)

Website

Get the PDF

We use your details to provide this resource and manage your request. Downloading does not subscribe you to marketing emails. Privacy policy

Talk through your CRM data and AI question with Nemely.