The TAM Build Execution Plan (7 steps)

Step 1: Audit the existing CRM (what do you already have?)

Before scraping anything, you need to understand what's already in the CRM. This is the same audit logic from our Audit article, but applied specifically to your TAM:

• Coverage: how many of your target accounts are already in the CRM? What percentage of your ICP is represented?

• Quality: of the accounts you have, how many are actually usable? (filled key fields, no duplicates, correct segmentation)

• Gaps: which segments, geographies, or verticals are missing entirely?

This step prevents the most common TAM mistake: importing accounts you already have, creating duplicates, and polluting dashboards.

A useful exercise: export your current account base, map it against your ICP criteria, and calculate the coverage gap. That gap is what the TAM build needs to fill.

Step 2: Define the TAM scope with the business

This is the strategic step most teams rush through. The TAM scope should answer:

• Who is your ICP? Not a vague persona: specific, measurable criteria. What industry, sub-industry, size indicators, geographic scope, and exclusion rules define a target account?

• What segments exist within the TAM? Not every account deserves the same treatment. Segment by potential value, strategic fit, or go-to-market motion (e.g., independent brokers vs. mid-market vs. enterprise).

• What data points do you need beyond firmographics? This is where differentiation happens. Depending on your business, the critical data might be: license types, regulatory registrations, technology stack, number of physical locations, recent fundraising, specific certifications, or hiring patterns.

The output is a TAM Scope Document that both the ops team and the business validate before any scraping begins.

The key insight: the more specific the data points, the more actionable the TAM. A TAM with industry + headcount is a starting point. A TAM with license type + regulatory status + distribution channel + revenue estimate is a targeting weapon.

Step 3: Identify data sources (this is where it gets hard)

Here's the truth: most valuable TAM data doesn't live in a single tool you can subscribe to.

For well-covered verticals (SaaS selling to other SaaS companies), tools like LinkedIn Sales Navigator, Apollo, or ZoomInfo give you reasonable coverage. But the moment your ICP is niche, regulated, or non-tech, the data you need is scattered.

Common data source categories:

• Public registries and government databases. Industry-specific: ORIAS for insurance brokers in France, BAFIN in Germany, DGSFP in Spain, FSMA in Belgium, IVASS in Italy. These are the official sources of truth for regulated industries. Similar registries exist in healthcare, finance, real estate, and other sectors.

• Association directories. Trade associations, professional bodies, and certification databases often have member lists that are more accurate than any commercial data provider.

• Web scraping and SERP research. For niche verticals, sometimes the best data comes from systematically crawling company websites, review sites (like Google Maps or Trustpilot), or running search queries designed to surface specific company types.

• Government APIs and open data. In France, data.gouv.fr and the Annuaire du Service Public. In other countries, equivalent open data portals. These give you unique identifiers (SIREN, SIRET, VAT numbers) that no commercial tool provides.

• Commercial data providers. LinkedIn Sales Navigator, Apollo, ZoomInfo, Clearbit, still useful, but as enrichment layers on top of your primary sources, not as the foundation.

The principle: start with the most authoritative source for your vertical, then layer enrichment on top. Don't start with a commercial provider and hope it covers your niche.

Step 4: Scrape a sample and validate with the client

This is the most important quality gate in the entire process, and the one that separates a good TAM build from a bad one.

Never scrape the full TAM in one shot. Instead:

1. Pull a limited sample (100–500 records) from your identified sources.

2. Enrich the sample with the data points defined in the scope.

3. Present it to the client team: "Is this the right data? Are these the right companies? Are the data points useful for your sales motion?"

4. Iterate based on feedback. Edge cases always surface at this stage: companies that look like they match the ICP but don't, data points that seem complete but aren't actionable, segments that need finer granularity.

This back-and-forth is where the TAM gets calibrated to reality. Skip it, and you'll import thousands of records that look clean but don't serve the business.

Step 5: Full TAM scrape and enrichment

Once the sample is validated, you run the full pipeline:

• Scrape from all identified sources across all target geographies.

• Standardize company names, industry codes, identifiers, and any custom fields. This is critical when working across multiple countries, the same concept (e.g., "insurance broker") may be categorized differently in each national registry.

• Enrich with additional data points: website, LinkedIn URL, revenue estimates, employee count, technology indicators, review scores, or whatever the scope document defines.

• Deduplicate against existing CRM records and within the scraped data itself.

The tooling varies by project. A typical stack: custom Python scripts for registry scraping, SERP research (via tools like HasData or custom APIs) for company validation, LLM-powered agents (on n8n or similar workflow platforms) for parsing unstructured data, and enrichment providers like FullEnrich or Apollo for contact-level data.

The important point: no single tool does this end-to-end. A serious TAM build is a data engineering operation, not a subscription.

Row-by-row enrichment matters. Iterating one record at a time catches errors that bulk enrichment misses, wrong industry classifications, subsidiary vs. parent confusion, outdated registrations. It's slower upfront but far more accurate at the end.

Step 6: Import into the CRM (the technical part everyone underestimates)

A clean TAM file is worthless if the import breaks the CRM. This step requires:

A) Field mapping

Every field in the TAM file maps to a specific CRM property. Before importing:

• Define which fields are new vs. existing.

• Set overwrite rules: does the TAM data override existing values, or only fill empty fields?

• Create dedicated import fields if needed (to preserve existing data during validation).

B) Deduplication logic

The import must check for existing records in the CRM. Matching criteria depend on the object:

• Companies: match by domain/website first, then by unique identifiers (SIREN, registration number), then by name + location as a fallback.

• Contacts: match by email first, then by LinkedIn URL, then by name + company.

Any near-duplicates should be flagged for manual review, not auto-merged.

C) Assignment and routing

Imported accounts need owners. Define the routing logic before import:

• Territory-based assignment (geography, segment, account tier)

• Round-robin within territories

• Ownership inheritance from existing relationships

D) Tagging and segmentation

Every imported record should carry:

• A TAM import tag (date + source, for traceability)

• Segment/tier classification

• ICP match score (if scoring is already in place)

Step 7: Sanity check + feedback loop

After import, run a structured quality check:

A) Verify the numbers

• Total records imported vs. expected

• Duplicate rate (how many were caught, how many slipped through)

• Fill rate on critical fields

• Assignment distribution (are accounts balanced across reps/territories?)

B) Validate with the team

The people who will use this data every day, SDRs, AEs, sales managers — are the final validation layer. But this only works with clear enablement:

• What was imported and what changed

• How to use the new segmentation fields

• Where to report data issues

• What the refresh cadence looks like

C) Build dashboards to monitor TAM health

Track over time: TAM coverage by segment, fill rate on key fields, account activation rate (how many imported accounts are being worked), and data freshness.

The feedback loop is the most underrated part of the process. A perfectly built TAM that sales never uses is wasted effort. A TAM is not a deliverable, it's infrastructure that needs to live in the CRM and get activated daily.