AI for sales in Business Central

Customers who stopped buying

The most useful sales question in a B2B business is rarely “who are our biggest customers” — everyone knows that. It is “who used to buy regularly and has quietly stopped”, because in a business with recurring orders churn is silent. Nobody cancels; they just stop ordering, and six months later someone notices.

The analysis needs a definition of normal per customer rather than a single global rule. A customer who orders every three weeks and has not ordered in eight is a signal. A customer who orders twice a year is not overdue at four months. That distinction is the difference between a useful list and one the sales team ignores after the second week.

• Customers with no posted document beyond their own typical interval
• Customers still ordering but at declining value or frequency
• Accounts that dropped specific product lines while continuing to buy others
• Prior-year revenue at risk, so the list is ranked by what it is worth

Discount leakage

Discount is the fastest way to lose margin without anyone deciding to. Business Central records what was invoiced against what the price list said, so the gap is measurable — by salesperson, customer, item and period.

What makes this analysis land is context rather than the number itself. A discount above policy on a large order that came with prompt payment may be exactly right. The same discount given routinely to a customer who also pays late is a different conversation. An assistant can put those two facts in the same answer; a discount report cannot, which is why discount reports get produced and then quietly ignored.

Performance and territory

Comparison against plan, between territories, between periods, by product mix. This is well-covered ground in Power BI and if you already have a dashboard that answers it, keep the dashboard — it is faster to read. Where AI adds something is the follow-up: the dashboard shows that one territory is down 14%, and the useful next question is why, which nobody built a visual for. Our note on Power BI and Business Central covers where each fits.

Working with the CRM data you actually have

An honest limitation. If your sales process lives in Business Central's built-in CRM, opportunity analysis works. If opportunities are tracked in Dynamics 365 Sales, another CRM, or a spreadsheet, then Business Central holds the outcome but not the pipeline — and an assistant reading it can analyse what was sold but not what was nearly sold or why a deal was lost.

This is fixable by integrating the other system, which is a larger project. It is worth knowing up front, because “why are we losing deals” is a common first request and is frequently unanswerable from ERP data alone. Our article on CRM or ERP first covers the underlying question.

Sales agents

Beyond analysis, a sales AI agent can prepare the follow-up: draft outreach for lapsed accounts, create quotes from a described requirement, flag orders that breach credit terms before they are released. Order release and pricing decisions typically stay with a person — margin is given away one approval at a time, and that is a decision worth keeping human.