Lead qualification

After a conversation ends, OyeChats analyses what the visitor said and scores each dimension of your chosen framework. Those scores combine into one composite score out of 100, and the composite maps to a tier. Nothing is asked of the visitor to make this happen. It is derived from the conversation they already had.

Pick one per chatbot under Advanced → Qualification. Each framework ships as an editable preset: you can change dimension weights, turn a dimension off, rewrite the answer options and their scores, and reorder the sequence the chatbot works through.

1. 1

   Each dimension is scored

   On its own scale, from the conversation content.

2. 2

   Scores are normalised and weighted

   Each dimension is expressed as a percentage of its own maximum, then weighted. Weights are relative, not required to sum to 100, and disabled dimensions are excluded from the maths entirely.

3. 3

   The result is a composite out of 100

   Which is what the tier thresholds are compared against.

Need      25
Timeline  25
Authority 25
Budget    25

Conversation order: Need → Timeline → Authority → Budget

Each dimension can optionally show the visitor a one-tap multiple-choice prompt ("When are you looking to get started?"). Those prompts are off by default for every dimension, because most B2B visitors read them as qualification fishing.

With them off, scoring still happens. It is inferred from the conversation text after the fact. You get the tier without interrogating the visitor. Turn individual prompts on if you want the faster, more explicit flow.

Alongside what the visitor said, OyeChats scores how they behaved, up to 20 points. This is intent evidence that never appears in the conversation.

Urgency ages. By default the timeline dimension loses 5 points and the need dimension 3 points per 30 days, so a lead that was hot two months ago does not keep sitting at the top of your list. Decay is configurable per framework and can be switched off.

Every scoring change is written to an append-only signal log with the dimension, the before and after values, and the source. Whether it came from analysing the conversation or from the visitor tapping a prompt. Open any lead to see that trail, and correct a score by hand if the analysis got it wrong.