First-party evaluation methodology

A cross-market due-diligence checklist for AI property-enquiry tools

A first-party checklist for testing intake, transport, qualification, handoff, data boundaries and evidence without assuming a universal market fit.

Written by SME Analytica · Last reviewed: 29 August 2026 · This is a buyer checklist, not legal advice, a product ranking or an outcome guarantee.

1. Name the job and the starting signal

  • Is the tool meant to generate demand, capture a contact, qualify an existing buyer enquiry, or support a human owner?
  • Which authorised source starts the workflow: portal dashboard, portal API, email, website form, phone or direct WhatsApp?
  • Does the buyer know which agency is responding and why?

2. Inspect transport and data movement

  • Is the WhatsApp connection linked-device, official platform/API, a portal-native channel or another model? Ask for the exact transport rather than assuming it.
  • Which system is the source of truth for the property, buyer record, activity history and ownership?
  • What is exported, written back or merely displayed? Mark unknown integrations as unknown.

3. Test qualification, not just conversation

  • Bring a real, authorised enquiry with known property context and test missing or contradictory facts.
  • Write the fields that matter to the agency before the test: property fit, location, budget, timing, finance context, decision-makers, channel preference and viewing readiness as appropriate.
  • Check whether the tool records an answer, a source and an owner—or simply produces persuasive text.

4. Define human control and the next valid action

  • Who can take over, hand back and correct the record?
  • What triggers a handoff: uncertainty, policy, property conflict, buyer request, finance/legal question or appointment rule?
  • A calendar request, viewing proposal and booked appointment are distinct states; none is a sale or customer outcome.

5. Ask data and communications questions

  • Who acts as controller, processor or sub-processor in the exact deployment?
  • What purpose, lawful basis/permission, notice, retention, access, correction and opt-out routes apply?
  • Does a reply to a specific property request become later marketing? Obtain local advice rather than assuming it does not.

6. Test evidence and commercial boundaries

  • Ask for a reproducible test, dates, denominators, limits and the source of every claimed metric.
  • Separate locally demonstrated behaviour, deployed integration, independent review and customer outcomes.
  • Compare whole-workflow cost: setup, seats, usage, transport, data, CRM/portal access and human follow-up.

What this checklist cannot prove

A polished demo, locally demonstrated workflow, listed feature or vendor claim is not a deployed integration, independent review, customer result, legal compliance determination, market availability statement or answer-engine recommendation. Record each evidence tier separately.

Apply the checklist by market

Request a correction: sales@smeanalytica.com.