Waterfall enrichment 2.0 has arrived#
Waterfall 1.0:
Manually export from tool → verify → import to CRM
Back in the day, before we had all these tools to automate waterfall enrichment, most sales or ops folks manually clicked through browser extensions—Hunter.io, Apollo, LinkedIn Sales Navigator—copy-pasting or importing anything that looked like a useful prospect into an email validator and moving those that passed the check into the CRM.
Needless to say, we were spending most of our time in these frustratingly inefficient processes. Especially, if you had to change some attribute in the export, or add another step, such as enriching phone numbers, you'd have to replicate all the previous steps to get to the result.
Tooling without orchestration
We've moved on from copy-paste hell—most tools now plug straight into your CRM or sit on top of LinkedIn. Yet the processes haven't caught up, so teams get stuck halfway between manual and automated.
Here's what I mean, I talk to SaaS scale-ups every week. They're packed with enrichment tools—Kasper, Apollo, Lusha, LeadIQ—which their reps cycle through until something works.
Each profile manually opened burns at least 60 seconds: Lusha → VoilaNorbert → SalesQL → Neverbounce/ZeroBounce/Scrubby (my personal workflow) to see if there's a valid email worth targeting.
What's often missing is the ability to knit these tools together to create a seamless experience for your sales team, who shouldn't need to waste their time selecting, enriching and validating these leads - but rather find creative way to talk to them.
Black-box waterfall providers
If you're using the same plug-and-play enrichment tool as everyone else, you're stuck with their blind spots—same stale data, same regional gaps, same logic you can't see or tweak. You can't layer in your own signals. You can't tune it to your market. And you definitely can't build an edge if everyone's running the same playbook.
Waterfall 2.0: Custom waterfall setups#
Before we see what good looks like, let's define what waterfall enrichment simply is.
Your goal is to maximize enrichment coverage for data points that matter to you at the lowest possible cost.
An email waterfall is basically an automated queue of providers fired in priority order until a field is filled and verified. This way you pay only for the first hit.
Lead → work_email? → mobile? → personal_email?
Apollo Apollo Apollo
Lusha Lusha Lusha
Cognism Cognism Cognism
The outcome: Doubles reachable prospects by chaining multiple data sources. Keeps the ICP tight by automatically using the most accurate vendor for each region.
Pitfall - Chasing completion over accuracy
Most waterfalls hit "done" the instant any provider hands back a value. Two problems:
- Completion ≠ precision: A bad phone number burns more rep time than an empty field.
- Bad data compounds problems downstream: One junk email tanks deliverability, dents domain reputation, and warps your metrics.
Rule of thumb: A false positive costs 3–5× more than a blank—it creates outreach, follow-ups, spam and cleanup you never needed.
How Descript built the right custom waterfall
Descript lifted data completion from 60% to 82% by running a 10-vendor enrichment cascade.
Here's how they did it:
- Sampled 100 verified leads per key country.
- Benchmarked each vendor—logged coverage (how many fields they filled) and accuracy (how many of those fills were right): They ran the identical set of 100 verified leads through each provider and captured whatever data they returned. They then checked those returns—numbers, emails, LinkedIn URLs to see what percentage were actually functional.
- Ranked vendors by accuracy → coverage → cost: Accuracy first, coverage second, cost last. Any vendor that couldn't clear a 90 percent accuracy bar was cut immediately, no matter how much ground they covered
- Bucketed vendors by region: e.g. Cognism + Lusha for EU, ZoomInfo + LeadIQ for US, etc.
- Monitored results weekly and iterated.
Do you need an LLM layer in your custom waterfall
A solid enrichment stack has layers. Vendor APIs give you the hard identifiers: emails, company names, phone numbers. LLMs are great for filling in the soft stuff—like whether someone recently raised a round, or what tools they're using. But LLMs can hallucinate, and vendors can go stale.
That's why the best setups combine both.
| Layer | Best For | Example |
|---|---|---|
| Vendor API | Hard IDs (email, DUNS, phone) | Clearbit → validated work email |
| LLM API | Soft facts (fundraise, tech) | GPT-4o fetches yesterday's TechCrunch round |
| Verification | Regex, SMTP, HLR | Cargo's built-in verifier |
Use vendors when you need structured, high-confidence data. Use LLMs when freshness matters or vendor doesn't cover public data that's really relevant to you. Then layer in verification—SMTP, regex, HLR—to make sure what you got is actually usable.
Cargo's waterfall engine
You don't need to stitch together vendors, logic, and verification pipelines on your own.
Here are the tools available to our users to start leveraging in their Plays.
| Engine | Primary Fields | Secret Sauce |
|---|---|---|
| Work Email | Work email + role data | 95% deliverability, no-bounce guarantee |
| Personal Email | Map PLG sign-ups to pro identity | Score filter keeps only High |
| Mobile | Direct dial or cell | Coupled with HLR ping |
We've fine-tuned it to prioritize both accuracy and completeness—with layered logic, fallback providers, scoring, and built-in verification.
| Step | Old Waterfall | Cargo 2025 Flow |
|---|---|---|
| Query vendor | ✔ | ✔ |
| Accept first hit | ✔ | ↘ only if confidence ≥ threshold |
| Verify (SMTP/HLR) | ✖ | ✔ |
| LLM fallback | ✖ | ✔ |
| Score multiple hits | ✖ | ✔ best score wins |
| Store provenance | ✖ | ✔ field-level tags |
Result: 100% golden-record rate with an audit trail RevOps can trust.
What to remember
Most teams are still stuck somewhere between manual enrichment loops and off-the-shelf tools that treat every business the same. But the game has changed.
We've moved from Waterfall 1.0 (manual copy-paste → tooling without orchestration → black-box vendor bundles)—and now, we're in Waterfall 2.0 (custom, layered enrichment stacks) built for accuracy, flexibility, and control.
Waterfall enrichment today means:
- Coverage is the floor, not the ceiling. The new KPI is golden-record rate.
- Accuracy beats presence. False positives erode ROI faster than blanks.
- API vendors + LLM is the winning combo. Deterministic IDs + real-time web intel.
- Cargo's trio (work email, personal email, mobile) + LLM fallback + scoring layer is the shortest path to complete, trustworthy records.
That's what Cargo's waterfall tools are built to do: verified work emails, mapped personal signups, enriched mobiles—all scored, verified, and tracked. No mystery logic. No stale hits. Just golden records, ready for action.
MaxMay 1, 2025
