A buyer writes on WhatsApp, “We will probably order next month.” A salesperson puts the full quote value into next month's forecast. Two weeks later, the buyer has not even secured budget approval. WhatsApp did not cause the error. The error was treating an encouraging message as evidence of a committed purchase.

The working rule: use conversations to gather evidence about an opportunity, but do not turn every positive reply into forecast revenue. A useful forecast needs a defined deal, defensible value, expected decision date and a confidence category the team can explain.

Interest is not commitment

“Do you offer this service?” establishes an inquiry. A quotation request shows evaluation, not approval. “We agree in principle” may still mean finance, management or legal review is pending. Even a purchase order needs a scope and terms check before you decide what is actually committed.

Microsoft's forecast category guidance separates early Pipeline, Best Case and Committed opportunities, while Won reflects a deal actually closed. That is a useful record-keeping model, not proof of a universal conversion probability for Egyptian companies.

What should a conversation contribute to the forecast?

Capture five reviewable facts: the legal entity making the purchase, the service and scope requested, quote value and validity, the person or team authorized to approve, and the next action with a date and internal owner. Keep a conversation reference in the appropriate system, but avoid copying unnecessary personal information into shared spreadsheets.

Above all, separate the customer's words from your team's inference. “Our director will review it on Sunday” is a reported fact. “The deal will close Sunday” is a forecast assumption that still needs evidence.

A weekly review that does not inflate the pipeline

Start with three operational groups. Qualification pending means important facts such as need or buying authority are missing. Under evaluation means there is an active offer and discussion but no final approval. Commercially committed means the authorized buyer has provided a clear commitment and your business has checked its required conditions. Record completed deals separately.

These are not ready-made probabilities. With enough historical closed deals, calculate stage conversion rates using consistent definitions, then check differences by sector and deal size. With limited data, show scenarios or ranges rather than a precise percentage that suggests certainty you do not have. Microsoft's forecasting guide likewise distinguishes Pipeline, Best Case, Committed and Won.

An illustrative example, not customer data

Imagine three quotes worth EGP 50,000 each. One buyer has only requested details. Another is waiting for procurement approval. The third has documented approval and an agreed implementation date. Reporting EGP 150,000 as “next month's sales” hides the difference. Report total pipeline value separately from a weighted forecast, explain the weighting assumptions, and do not recognize any of it as actual revenue until the relevant conditions are met. These figures are entirely hypothetical.

When the buyer changes the plan

A new decision date, purchasing entity, service requirement or budget is not just another chat note. Revisit the opportunity value, stage and expected close date. A buyer who has stopped responding should not remain in Committed without evidence. If two subsidiaries have separate purchasing decisions, do not double-count one group-wide deal. See our guide to parent and subsidiary opportunities.

Where an AI assistant helps, and where it stops

According to Mr. AI's published product information, the platform supports knowledge-based replies, follow-up, conversation summaries and human handover. Those can help a salesperson recover the promise and next step. They do not establish that Mr. AI produces approved financial forecasts or automatically creates CRM opportunities. Forecasting still requires a human decision and a reliable deal register.

Review your last twenty WhatsApp-sourced opportunities. How many have an identified decision-maker? How much forecast value rests on a phrase such as “hopefully next month”? Fix stage definitions first. Then try organizing customer conversations or book a discussion if follow-up context is being lost.