Who: I work in a cross functional role as a web developer and marketing specialist. My primary goal is to leverage all the technologies I have available to convert more potential clients into engaged clients.
Goal: Find a flexible and reliable way to grade conversion potential of contacts post iOS 14.5. Apple’s privacy policies are only getting more strict, and formerly reliable tactics are more or less useless now. My hope is to leverage an LLM to process the language included in our “message” contact property. If the message property in their form submission shows some from of buyer intent, the contact can be appropriately nurtured. If language processing can reliably indentify buyer intent, or at least filter out unqualified contacts, it could be a significant step in the right direction.
Examples: As far as I know, other CRMs do not currently offer this level of AI integration.
Screenshots, pictures, etc: My institution’s data privacy policies prevent me from sharing customer data in this way, but I can share an outline of my test process. I ran a test on a few contacts that I manually qualified to see if an LLM would come to the same conclusions. For this purpose, I used claude. I sent a name (I used aliased names like John Smith to protect contact identities), as well as the messages they sent, and asked claude to categorize the contacts as unqualified, or potentially qualified. Claude successfully categorized the contacts in the same way I did. My assumption is that GPT-3.5 which Breeze runs on would be comparably capable. As an addition, I asked for a score out of 100 for “potential to close” and it provided reasonable estimates that could be used as a basis for marketing follow up.
If this capability was added, users could potentially run workflows nightly, weekly, or monthly to categorize leads based on the content of their messages.