If you have used a single data provider for emails, you know the feeling. It finds 60% of them. The other 40% are blanks, and the blanks are not random. They are often the smaller, newer, more interesting companies.
A waterfall is the fix for that. It is also one of the easiest ways to burn through Clay credits, so it is worth understanding before you add one.
How a waterfall works
You pick a field you want, say a work email, or the latest funding round. Clay lists several providers that can supply it, in an order you control.
Then it runs like this for each row:
- Ask provider one. If it returns a result, stop.
- If not, ask provider two. If it returns a result, stop.
- Keep going down the list until something answers or the list runs out.
Because it stops at the first hit, you are not paying for every provider on every row. You pay for the ones that actually returned something, and Clay's own description says you are not charged for most providers that come back empty.
A real example
In a table I built to find recently funded software companies, I added a "latest funding round" waterfall. It checked four providers in order.
Here is what happened across the first few companies:
- One company answered from the first provider, and the rest showed "run condition not met", meaning they never ran.
- Another needed the second provider before it got funding data.
- Others went further down the list.
The funding amounts that came back ranged from about a million dollars to tens of millions. Each result came from whichever provider happened to know that company, and no single one would have found all of them. That is the whole argument for a waterfall in one table.
What it costs
The estimate Clay showed for that funding waterfall was about 5.8 credits per row, up to 12 if it ran through the whole list. Compare that with about 0.1 credit for an AI column on a small model.
So the question is not "is a waterfall good". It is "is this field worth six credits a row".
When it is worth it
| Field | Worth a waterfall? |
|---|---|
| Work email on a list you will actually send to | Yes. Bad emails cost you deliverability |
| Funding round you will mention in the first line | Often yes |
| Employee count | Usually no, one provider is enough |
| Industry or description | No |
The rule I use: pay for a waterfall when a blank costs you more than the lookup does.
How to keep the cost down
- Filter first. Run cheap columns that remove bad-fit companies, then waterfall only the rows that survive.
- Order providers by price and coverage. Put the cheaper, broader sources first.
- Add a run condition so the waterfall skips rows you have already disqualified.
- Test on three rows and read the estimate before running a table. I go through all of this in how to stop burning Clay credits.
Do not forget to verify
A waterfall finds an email. It does not prove the email is safe to send to. Run a verification step after it, and send only to addresses that pass. Otherwise you have spent credits to find addresses that hurt your sending reputation, which is the problem I describe in why cold email lands in spam.
The short version
A waterfall is a list of providers tried in order until one answers. It raises your hit rate and it costs real credits. Use it on the fields that matter, filter before it, and verify after it.
If you are deciding whether a waterfall should live in a table or in a workflow, see Clay Workflows vs tables.