Check commercial HVAC CRM data before an import

Start with the destination template

An AI-generated tidy spreadsheet is not necessarily a valid CRM import. Obtain the current destination template and instructions from the administrator or implementation team. Use a permitted sample first, keeping the original export unchanged. Record the source file version, destination template version and the entity being imported: customers, locations, equipment, jobs or estimates.

ServiceTitan documents separate import templates and required versus optional columns. Its onboarding guidance directs customers to an implementation manager for templates and assistance. Do not assume that a generic CSV, a different vendor’s template or an old export can be uploaded safely. This workflow prepares review artifacts; it does not perform a migration or grant import permissions.

Build a field map before transforming rows

Use columns for source field, destination field, entity, required status, allowed transformation, example and unresolved question. Keep customer billing data distinct from service-location data. Preserve source IDs as text where appropriate so leading zeros or long identifiers are not accidentally changed by spreadsheet formatting.

For a fictional customer sample, one row has a billing street but no service-location street. If the destination requires the latter, put the row in an exception report. Do not copy the billing address into it without evidence. An absent optional contact field should stay blank under the approved rules; a fake email address can cause real communication and record-association problems later.

Reconcile every row before authorization

Ask AI to flag missing required fields, unexpected values, ambiguous dates, duplicate source IDs and unapproved transformations. Return a row number and reason for each exception. Use deterministic spreadsheet checks for counts and required fields as well; do not rely solely on a language model to notice every malformed row.

Compare total input rows with approved rows plus rejected rows and account for any deliberate split. Have the administrator approve the mapping and exception decisions before a controlled import through the official process. Retain the original, reviewed mapping and results. After the authorized import, reconcile record IDs and a representative sample in the destination. A successful file upload alone does not prove that parent/location relationships or historical associations were preserved.

Review checklist

• Obtain the current official template and import owner.
• Preserve the original export and stable source IDs.
• Document transformations before applying them.
• Reject missing required values without inventing replacements.
• Reconcile counts and inspect destination relationships after import.

Source context and limits

ServiceTitan supports the template and required-field context; its customer/location documentation also warns against placeholder contact information. Mapping, reconciliation and exception review are our proposed controls, not a tested migration or a claim of native AI import validation.