Waterfall Enrichment Explained: How B2B Contact Data Actually Gets Verified
What Is Waterfall Enrichment? (And What It Isn't)
Waterfall enrichment is a B2B data method that queries several providers in sequence for the same field, stopping at the first one that returns a result. Ask for a prospect's mobile number, and the request hits the first provider in your chain. If that comes back empty, it falls through to the second, then the third, until something lands or the chain ends.
It has nothing to do with the waterfall model in software project management. Same word, unrelated discipline.
A waterfall is a routing method, not a database, and vendors sit in three different positions:
- Aggregators orchestrate other companies' data rather than owning any. Clay sells access to "150+ databases"; BetterContact and FullEnrich each aggregate 20+ vendors. None markets a first-party contact database.
- Hybrids run their own database and layer a waterfall on top. Apollo is explicit about the order: third-party sources are checked only "if Apollo doesn't have the contact information you need."
- Single-source providers are not waterfalls, though they appear inside almost everyone else's chain. Hunter.io, ContactOut, and People Data Labs each sell one dataset rather than routing between others'.
A waterfall describes how sources are queried. Confirm whether your provider orchestrates other datasets, owns a dataset, or combines both.
How Waterfall Enrichment Actually Works
Take 1,000 accounts with a first name, last name, and company domain, where you need a work email and a mobile number for each.
In one illustrative Apollo-first configuration, the work-email chain runs like this. Apollo answers on records inside its own database, which it always checks first in its own product. The misses fall through to Hunter, which crawls the public web. What Hunter misses falls to People Data Labs, which holds licensed rather than crawled supply and catches a different slice. The order after the first provider is yours to set: an Apollo admin chooses which connected third-party sources run, and in what sequence.
Anything still empty at the end stays empty. Whether the misses cost you depends on the provider: Hunter notes that some charge only on a result while others charge per lookup regardless, and Apollo consumes a credit whenever a source returns an email, verified or not.
The mobile chain is a separate sequence with a separate roster, and that is not an implementation detail. Clay names part of each roster: nine providers for work email and three for mobile, both lists ending in "& more," and People Data Labs is the only name appearing on both. A vendor strong on corporate email addresses has no particular reason to be strong on personal cell numbers, because the two are acquired in completely different ways.
Not everyone routes sequentially. ZoomInfo says its GTM Studio queries vendors "in parallel rather than sequentially," scoring for the highest-confidence match "not just the first one found." That trades the waterfall's cost discipline for match quality, and it is also ZoomInfo's argument for buying ZoomInfo.
Three rules govern the sequential version:
1\. Order decides whose price you pay. The chain stops at whichever provider returns data, so put the best expected cost per successful match first.
2\. Check the billing model before assuming a miss is free. Some providers bill only on a result, others bill per lookup, and a returned record is not necessarily verified.
3\. One chain per data type. Work email, personal email, and mobile each need their own sequence.
Why Sequence (Not Just Stack) Data Providers
Every explainer asserts that providers hold different records. Almost none says why, and the why is what tells you how to order the chain. Providers differ not because some are better, but because they acquire data in structurally incompatible ways.
| Provider | How it acquires data | What that means for coverage |
|---|---|---|
| Hunter.io | Crawls the public web only | Strong where an address was published somewhere indexable. Hunter's product line is email only, with no phone product listed. |
| ContactOut | Runs its own crawling infrastructure across the open internet | ContactOut publishes an unusual ratio: 150M personal emails against 200M work. |
| People Data Labs | Licenses commercial supply from HR tech, real estate tech, identity and anti-fraud vendors, plus public records | Reaches records that were never published publicly. Sells to other waterfalls rather than running one. |
| ZoomInfo | Proprietary technology, machine learning, public sources and a contributory network | Its contributory network and human research reach records no crawler can see; its public-source component overlaps with the crawlers. |
Three acquisition models, three different universes of records. That is why chaining them adds coverage, and why chaining three crawlers adds little.
Freshness compounds it. Hunter removes contact information after six months when its public source disappears, so a record correct in March can be legitimately absent in October. Providers prune on different schedules, so a chain buys coverage across time as well as across sources.
What Waterfall Enrichment Can Verify
Contact-data waterfalls commonly target work email, personal email, and mobile. The same routing pattern also runs on firmographic, technographic, funding, and intent fields, so treat waterfall enrichment as a method rather than a fixed list of outputs. On the contact side, the three fields are not equally available. People Data Labs publishes exact record counts for its own datasets: 2.47 billion profiles in total, 778.9 million with a phone number of any kind, 611.8 million with an email address and 478.4 million with a mobile number. Mobile is the scarcest. That is what you would expect from a field rarely published and mostly licensed, though inside one identity graph the gap against email is about 1.3x rather than an order of magnitude.
Coverage depends on the requested field and target segment. Adding more providers does not guarantee a higher verified match rate; evaluate the incremental results on the same sample.
Waterfall Enrichment vs. CRM Enrichment
These get conflated constantly, partly because several vendors sell both motions from one subscription. Apollo is the clearest case: its waterfall only fires when Apollo's own database comes up empty.
CRM enrichment describes where records are updated; waterfall enrichment describes how providers are queried. A waterfall can refresh existing CRM records or add details to newly discovered prospects. In either case, records need periodic revalidation.
The Real Benefits (Coverage, Deliverability, Cost Control)
Coverage. Each provider contributes the slice its acquisition method reaches, so fill rate rises. Every vendor page leads with this one, and its size varies most by segment.
Deliverability. The underrated one. A chain configured to require verified results, or followed by a validator, keeps bad records out of your sequencer, and hard bounces are among the fastest ways to damage a sending reputation. Note the caveat: Apollo can be set to stop on any returned email, verified or not, so verification is a setting rather than a guarantee. On the same two months of beta-customer data behind its coverage figures, Apollo reports a 45% lower email bounce rate from turning waterfall enrichment on. A missing row costs one prospect. A hard bounce costs a fraction of every future send from that domain.
Cost control. On hit-based providers, you are billed for matches rather than attempts, so a provider that misses on your segment is cheap to keep. On lookup-based providers, it is not, which is why rule two matters before adding a fourth source.
When Waterfall Enrichment Isn't Worth It
The value of a waterfall depends on the starting dataset and the segment you need to reach.
| Claim | Baseline it is measured against | Source | What they sell |
|---|---|---|---|
| Waterfall "routinely triples our customers' data coverage and quality" | Baseline unstated; a customer replacing whatever they had | Clay's waterfall page | Access to 150+ databases |
| 5% more email coverage, 7% more phone numbers | Incremental, on top of Apollo's own database | Apollo's waterfall product page, from two months of beta-customer data | A waterfall layer over its own database |
These vendor-reported figures use different baselines and cannot be compared directly. If the first provider covers little of your list, additional sources can add substantial coverage. If it already covers most records, the remaining sources may contribute only a small increment.
Choose providers by the additional verified coverage they deliver on your sample. Access to a large catalog does not mean every source should be queried for every record. Add a provider only when its incremental matches justify the cost and latency.
Skip the waterfall, or keep it to two providers, when:
- Your list is small. A lightweight lookup process may cost less to operate than a multi-provider integration. Compare the time saved with setup, usage and maintenance costs.
- A single good source already covers most of your list. Cleanlist's 500-lead benchmark puts single-source at 70 to 80% verified email against 98% for its own 25-provider waterfall. Cleanlist sells the waterfall, so read the single-source figure as a floor. A good single source already reaches roughly three-quarters of a well-covered list, and the chain buys the rest at full price.
- Your buyers are not corporate. Every provider in the chain assumes a buyer with a corporate digital footprint. Local trades, sole proprietors, and offline businesses fall outside that assumption, and stacking more providers does not fix it.
- You cannot act on the extra rows. Coverage you never use is a line item, not an asset.
And one thing nobody publishes: coverage broken down by seniority. No provider releases match rates by job level, and no independent benchmark of the category exists in public. Every number above came from a company with something to sell.
How We Use Waterfall Enrichment in Prospector
Synter Prospector uses waterfall discovery with signals such as technology use, hiring patterns, and demographics. Enrichment can run when a signal fires, reducing the delay between lookup and outreach. The freshness of a returned record still depends on the underlying provider.
What happens after the match is what most enrichment tooling leaves to you. Prospector routes verified contacts into dedicated secondary sending domains with automated SPF, DKIM, and DMARC, so the system that enriched the record also provisioned the domain it will be contacted from. Over a 90-day benchmark, that pipeline held inbox placement at 99.4%. That is our own number from running the engine on our own outbound, self-reported and not independently audited. If you are weighing this against a hybrid provider's built-in waterfall, the Apollo comparison covers the differences.
Contact our team to discuss enrichment and audience activation.
FAQ
What is waterfall enrichment?
It fills in B2B contact data by querying multiple providers one after another for the same field. The chain stops at the first provider that returns a result, and each one catches records the previous one missed. Whether that result is verified, and whether a miss costs a credit, depends on the provider.
Is waterfall enrichment the same as CRM enrichment?
They describe different dimensions. CRM enrichment updates records in a CRM; waterfall enrichment describes querying providers in sequence. You can use a waterfall to enrich existing CRM records or newly discovered prospects.
What does waterfall enrichment mean in Apollo?
Apollo checks its own database first and only falls through to your connected third-party sources, in an order you define, when it has no match. You can turn the third-party layer off entirely, which makes Apollo a hybrid rather than a pure waterfall.
What's the difference between a waterfall and one big database?
A single database returns only what its own acquisition method captured. Sequencing across a crawler, a contributory network, and a licensed broker reaches records none of them could alone.
How many providers should be in a chain?
The vendors themselves suggest two to three (Hunter) or two to five (ZoomInfo). Beyond that, marginal coverage tends to shrink while hit, or lookup costs and orchestration overhead keep rising. Benchmark lift, verification rate, and bounce rate on your own segment before adding another source.
Can you turn off waterfall enrichment?
Yes, in the tools that layer it over their own data. Apollo lets an admin remove third-party sources from the enrichment lineup entirely and run Apollo-only. In pure aggregators, the waterfall is the product, so the equivalent control is trimming the provider list rather than switching the method off.