AI-Qualified Lead Generation for a Commercial Fitness Supplier

We designed three independent Make scenarios to handle business discovery, company enrichment, and AI-assisted qualification.

Google Sheets serves as a shared staging layer, allowing each scenario to process records independently while preserving the information generated at previous stages.

Apollo provides company and contact research, while Claude evaluates businesses against the client's qualification criteria. Explicit workflow rules determine which records proceed to HubSpot and how existing companies and contacts are matched.

This architecture separates AI-assisted decision-making from CRM operations, making the system easier to maintain and reducing the risk of irrelevant or duplicate records.

Scenario 1 — Business discoveryAn Apify dataset notification supplies the listing data.

The collection scenario preserves business names, categories, addresses, phone numbers, websites, Maps URLs, and source place IDs. A sheet lookup and zero-result gate precede insertion, keeping record checks separate from collection.

The business listings are stored in Google Sheets. The original listing remains available as the starting point for later research.

Scenario 2 — Company enrichmentThe enrichment scenario reads unprocessed business records from Google Sheets.

JavaScript trims values, normalizes website hostnames, removes the www prefix, and groups records by domain. Businesses without usable websites are routed to a separate sheet rather than sent into domain-based enrichment requests.

Apollo returns available company context, including industry, employee estimates, location, description, and keywords. Response matching associates organization data with the input businesses using exact domains, a root-domain approximation, and name-based fallbacks.

The resulting company information is written back to the Google Sheets staging layer for qualification.

Scenario 3 — AI qualification & CRMClaude receives listing and company evidence alongside the supplier's target-facility criteria, including relevance to indoor cycling programs.

It returns structured fields for target fit, model-generated confidence, facility type, indoor cycling activity, supporting evidence, estimated bike brands, product opportunities, priority, and research notes.

The workflow parses those fields, records the assessment in Google Sheets, and gates company processing on a positive target decision. HubSpot company handling searches by domain before creation or updates.

Facility type informs Apollo contact research and role criteria. Email-bearing contacts are looked up by email before create/update actions; new contacts receive company associations.

1. Separate processing responsibilitiesDiscovery, enrichment, and qualification operate as independent scenarios. This keeps responsibilities distinct and makes each processing stage easier to inspect and maintain.
2. Use appropriate identity matchingSource place IDs retain business-listing identity. Normalized domains support company enrichment and HubSpot company matching. Email addresses support contact matching. These identifiers should not be treated as interchangeable, and email matching applies to contacts with an available email address.
3. Handle incomplete company informationBusinesses without usable websites are routed separately instead of being sent into domain-based enrichment requests. This preserves incomplete records without introducing avoidable API errors.
4. Keep AI decisions explainableClaude returns structured qualification information with supporting evidence and research notes. The model-generated confidence score is an assessment, not a statistically calibrated probability.
5. Protect CRM data qualityAI qualification is separated from HubSpot record operations. Explicit workflow logic controls qualification gates, company lookups, contact matching, and record synchronization.

The implemented system connects business discovery, enrichment, qualification, and CRM synchronization into a structured prospecting process.

Businesses are evaluated against the target customer profile before entering HubSpot. Existing company and contact records are checked before new records are created, while incomplete source records remain available for further processing.

The result is a more controlled lead generation process with clearer qualification evidence and less reliance on manual data handling.

• Structured qualification

  Business relevance is assessed using enriched company information and target customer criteria.

• CRM data quality

  Company and contact lookups are performed before record creation or updates.

• Traceable prospect research

  Source listings, enrichment data, and qualification results remain available through the staging workflow.

• Controlled AI integration

  AI assessments are separated from CRM synchronization logic.

Make

Workflow orchestration across the three scenarios.

Apify

Business listing collection.

Google Sheets

Shared staging, processing handoffs, and qualification records.

Apollo

Company enrichment and contact research.

Claude

Structured AI-assisted qualification.

HubSpot

Company and contact management.

JavaScript and HTTP APIs

Data normalization, response matching, and integration logic.