Enrichment Waterfall Strategy for GTM Teams
Single-vendor enrichment tops out around 40-60% match rates. A properly sequenced waterfall hits 85%+. Here's how to build one.
Why Single-Vendor Enrichment Fails
Every enrichment vendor has coverage gaps. ZoomInfo dominates enterprise contacts but struggles with startups under 50 employees. Apollo has strong email coverage but weaker direct dials. Clearbit excels at firmographics but trails on personal email addresses. No single provider covers every segment of the B2B market with equal depth.
The numbers bear this out. Run a list of 1,000 mid-market companies through any single enrichment provider and you'll get back 400-600 complete records. That leaves 40-60% of your list with missing fields: no direct email, no phone number, incomplete firmographic data, or stale information that bounced two quarters ago.
GTM Engineers solve this with waterfalls. Rather than accepting one vendor's partial coverage, you sequence multiple providers so each one fills gaps the previous one missed. The concept comes from lead routing (where leads "fall" to the next handler if the first doesn't pick up), but the data enrichment version is more structured. Each tier in the waterfall has a specific purpose: primary coverage, gap filling, verification, or specialty lookup.
Waterfall Architecture
A production waterfall typically has three tiers. Each tier processes only the records that the previous tier left incomplete.
Tier 1: Primary provider (broadest coverage). This handles the bulk of your enrichment volume. Choose the provider with the best match rate for your target market. For enterprise B2B, that's usually ZoomInfo. For SMB and startup targets, Apollo often matches more records at a lower price point. Clay's enrichment layer can serve as Tier 1 because it aggregates multiple sources in a single call, though this reduces your control over provider sequencing.
Tier 2: Gap filler (different data source). Records that Tier 1 missed or partially filled flow to Tier 2. The key is choosing a provider with a different underlying data source. If your Tier 1 is ZoomInfo (web crawling + partnerships), your Tier 2 should pull from a different corpus. Clearbit (firmographic APIs), Lusha (social network data), or FullEnrich (multi-source aggregation) each have distinct data lineages that complement ZoomInfo's coverage. Tier 2 typically catches 30-50% of the records Tier 1 missed.
Tier 3: Specialty lookup (targeted fills). After two broad passes, the remaining gaps are usually specific fields: direct dials, personal emails, or niche firmographic data. Tier 3 uses specialized tools. Cognism for European phone numbers. Hunter.io for email pattern discovery. LinkedIn Sales Navigator for title verification. This tier is the most expensive per-record but processes the smallest volume.
Cost Optimization Strategies
The naive approach to waterfall ordering is "most accurate first." Start with the best provider, fill gaps with the second-best, finish with the cheapest. This maximizes data quality but wastes money. Your most expensive provider processes every record, including the easy ones that any provider could match.
The cost-optimized approach flips this: cheapest first, most expensive last. Run your list through Apollo ($0.01-0.03/record) before ZoomInfo ($0.15-0.50/record). Apollo will match 50-60% of records at a fraction of the cost. ZoomInfo only processes the 40-50% that Apollo missed, cutting your ZoomInfo spend in half.
The tradeoff is accuracy. Cheapest-first waterfalls may return lower-quality data for records that both providers can match (the cheaper provider's data might be older or less verified). The practical compromise: use cheapest-first for high-volume prospecting lists where some data staleness is acceptable, and most-accurate-first for priority accounts where data quality directly impacts conversion rates.
Credit management matters. Most enrichment APIs charge per successful lookup, not per attempt. Structure your API calls to check if a record already has the field you need before making an enrichment call. If Tier 1 returned a valid email, don't send that record to Tier 2 for email enrichment. This sounds obvious, but without explicit field-level routing logic, your waterfall will re-enrich already-complete records and burn credits.
Accuracy Benchmarks
Vendor-reported accuracy numbers are marketing. Real-world accuracy depends on your target market, record age, and what fields you're enriching. According to ZoomInfo's own data quality benchmarks, business email accuracy runs 85-95% for enterprise contacts and drops to 70-80% for SMB. Phone number accuracy is consistently 10-15 points lower than email across all vendors.
Here's what we've observed across production waterfall pipelines.
Business email accuracy: ZoomInfo 88-93%, Apollo 82-87%, Clearbit 80-85%, Lusha 78-83%. These ranges account for market segment variation. Enterprise contacts cluster near the top of each range; startup contacts near the bottom.
Direct dial accuracy: ZoomInfo 70-78%, Cognism 72-80% (European numbers), Apollo 60-68%, Lusha 65-72%. Direct dials decay faster than email addresses. A phone number verified six months ago has a 20-30% chance of being stale.
Firmographic accuracy: Clearbit 90-95%, ZoomInfo 88-92%, Apollo 82-88%. Company size, industry, and funding data are more stable than contact data, so accuracy rates are higher across the board.
Implementation with Clay
Clay is purpose-built for enrichment waterfalls. Its "Enrich" column type lets you chain multiple providers in sequence, with conditional logic that routes records based on which fields are still missing. This is the fastest path to a working waterfall.
The Clay waterfall setup: Create a table with your prospect list. Add an enrichment column using your Tier 1 provider. Add a second enrichment column using Tier 2, with a condition: "Only run if [email] is empty." Add Tier 3 with similar conditions for remaining gaps. Clay processes each row through the waterfall automatically, stopping when all required fields are filled.
The limitation: Clay's pricing scales with row volume and enrichment credits. At 10,000+ records per month, a custom Python waterfall that calls provider APIs directly becomes more cost-effective. Clay charges for its orchestration layer on top of the enrichment provider costs. For smaller volumes (under 5,000 records/month), Clay's convenience outweighs the markup.
For teams building custom waterfalls without Clay, the API integration patterns guide covers the Python scripting approach: API client setup, rate limit handling, retry logic, and result caching.
Measuring Waterfall Performance
Track four metrics to know if your waterfall is working.
Overall match rate. The percentage of input records where at least one required field was successfully enriched. Target: 85%+ for business email, 65%+ for direct dials.
Per-tier contribution. What percentage of total matches came from each tier? A healthy distribution: Tier 1 handles 55-65%, Tier 2 handles 25-30%, Tier 3 handles 10-15%. If Tier 1 handles less than 50%, your primary provider may not be the right fit for your target market.
Cost per enriched record. Total enrichment spend divided by total successfully enriched records. This number tells you whether your cheapest-first vs most-accurate-first ordering decision is paying off. Benchmark: $0.05-0.15 per enriched record for a three-tier waterfall.
Data freshness score. Sample 100 enriched records monthly and verify accuracy manually (check LinkedIn profiles, call phone numbers, send test emails). A freshness score below 80% means you need to re-enrich your database more frequently or add a verification tier to your waterfall.
The data enrichment tools category page lists all the providers mentioned here with detailed reviews and pricing comparisons.
Source: State of GTM Engineering Report 2026 (n=228). Salary data combines survey responses from 228 GTM Engineers across 32 countries with analysis of 3,342 job postings.