One buyer, three conversations: how many leads do you have?
A customer asks about a course on Messenger, returns through website chat, then follows up on WhatsApp. Counting each conversation as a new lead inflates the pipeline. But automatically merging people with similar names can be worse: it can expose information to the wrong person. The safer approach is to separate a person, a conversation, and a sales opportunity, and require evidence before joining identities.
Three records that should not be confused
A person is the individual or organization you have a relationship with. A conversation is a channel-specific interaction at a particular time. An opportunity is a commercial need with an owner, stage, and next action. One person may have several conversations about one purchase, or two genuinely different opportunities.
Your reporting should keep these levels separate. Otherwise a rising number of conversations may be mistaken for an increase in distinct prospective buyers.
When is it safe to link two records?
A phone number can help identify a WhatsApp contact within a known workflow, but it does not automatically prove that a Messenger profile belongs to that person. Matching names and similar questions are weak signals.
Use three outcomes in your review process:
Confirmed match. The customer explicitly confirms both accounts or your verified sign-in process supplies a reliable identifier. Link the person records while preserving each conversation's channel and original source.
Possible match. Similar name, timing, or request details, but no reliable confirmation. Flag the records for human review rather than merging automatically.
Do not match. The records conflict, or the customer asks you not to connect them. Keep them separate.
Identity matching and marketing permission are different questions. Confirming that two conversations involve the same person does not transfer consent across channels or purposes.
Example: a training center
A learner asks about a course on Messenger, fills in a website form with a phone number, then asks on WhatsApp whether another start date is available. The sales team initially sees three leads. After the learner confirms the link, the team can retain three conversations, one person record, and one course opportunity if the purchase need is unchanged.
Now consider a parent sharing a phone number while asking about a different student. Automatic merging would risk mixing two people's records and sending the wrong follow-up.
A practical one-week review
Start with 30 suspected duplicate records as an illustrative pilot, not an industry benchmark. Record the matching evidence, confidence level, reviewer, and decision. Preserve original source, conversation owner, and next action. Track how many suggested matches were confirmed and how many merges had to be reversed.
Do not interpret a fall in raw lead count after deduplication as lost sales. The goal is a reliable pipeline, not the smallest possible database.
What Mr. AI does and does not claim
The current Mr. AI FAQ confirms support for WhatsApp, Messenger and website chat, along with company-knowledge replies, summaries, and follow-up. It does not establish that Mr. AI automatically merges customer identities across channels. Apply this framework through your CRM or a human review process, and verify any specific integration before relying on it.
For the related but different problem of preserving what the customer already said, read our cross-channel conversation context guide. To assess your messaging workflow, try Mr. AI with non-sensitive test records.
Method note: This is an operational decision framework, not a measured Egyptian study or a claim about an automatic product feature.
