The Ultimate Clay.run Tutorial: Build an AI-Powered B2B GTM Engine

BizatWeb
67 followers
15 Video Views·Jun 9, 2026
Build a modern B2B GTM strategy. A guide for Sales & RevOps to learn Clay's AI, Waterfalls, and Intent Signals.
Want to get clients using cold email?
Start here 👇
https://app.outboundest.com/r3emGnyc
Use code: OUTBOUND (20% OFF)
🔥 Get the exact tools I use:
👉 https://bit.ly/bizattools
🚀 Grow Your Business & Master Digital Skills with BizatWeb
🌐 Website:
https://bizatweb.com
📩 Email:
[email protected]
📱 WhatsApp:
https://wa.me/212637648713
📲 Follow BizatWeb:
Instagram → https://instagram.com/BizatWebHQ
Facebook → https://facebook.com/BizatWebHQ
TikTok → https://tiktok.com/@BizatWeb
🔥 Business • Marketing • AI • Automation • Cybersecurity • Online Growth
Clay Basics and the PETE Framework. Hey there, so I am Guilherme and you can call me Gui,
it's a lot easier and welcome to chapter one of this clay course. Before we dive into automations,
workflows or the shiny stuff, let's get grounded. So this chapter is all about foundations.
So what clay actually is, what it's best used for and how to start your projects the right way.
So if you have ever felt overwhelmed by a blank clay table, this will help you get unstuck
and build with intention. Let's dive in. So what is clay? So at its core, clay is a data
workspace. AirTable, Zapier and Apollo had a super smart baby. You can pull in data from
hundreds of sources, enrich it, filter it, score it and push it wherever you need, like instantly
or any other outreach tool. But the key is this, clay isn't just a list tool, it's a thinking tool,
right? So the most common use cases, the first one of course is building lists of leads,
either people or companies enriched with all kinds of data points. But it's also great for
things like monitoring target accounts or surfacing high intent buyers or building outbound workflows.
So if you're in sales, rev ops or growth, this is your sandbox. Clay is insanely flexible.
That's both its magic and its challenge. So when you can't do anything, it's very hard to know
what to do first. This is a common problem in creative work. Writers, musicians, designers all
hit the same wall, which is a blank page, right? That's too many choices, no clear starting point.
We call it the blank canvas problem. The way out is structure. It's a simple framework that gives
you a way to start thinking, not just building, right? That's what we are going to talk more about
right now. Writers have outlines and musicians have scales. In clay, we use a framework called
FEAT. So let's call it FEAT to make things easier. The framework won't do the thinking for you,
but it will give you a way to start making sense of complex ideas. So let's dive into FEAT's
framework. This framework is the core mental model that will help you solve any list building
problem in clay. Think of FEAT as making a recipe instead of just throwing ingredients in a bowl.
So it stands for find, enrich, transform, and export. So let's break that down. Every clay project
starts with the find step. This is where you decide where is the raw data coming from. So we have a few
options. First-party data, third-party data, or intent data. Let's go over each one of those. So
first-party data, you can pull contacts, companies, or custom objects directly from your CRM or your
data warehouse. So imagine you have HubSpot, you are pulling contacts or companies out of there
and making use of it, enriching it, and going back to the CRM. The third-party data, so clay
lets you build net new lists from sources like clay, native, find people, native, find companies, or
for example, jobs, data sets, or also external sources like, which is a good use case, Google
Maps, Phantom Buster, Apify. So you can scrape businesses from Google Maps. It's very helpful.
And the last one is intent data. So we can start dynamically by monitoring, for example,
job changers, or live site visitors, or trigger events to build a list that updates in real time.
So this is your foundation, guys. So get this part right and the rest flows more easily, okay?
Now the second, enrich, which is to add the right contacts. So once you have found your database,
it's time to enrich. That means filling the missing contacts about each person or company
so you actually know why and how to reach out. You can, for example, waterfall enrichments. You can
use over 100 native providers, and this also helps you balance data coverage with cost, right? So
instead of, for example, paying Apollo, or paying Findme, or paying Hunter, and paying each one of
those, we can just use their native integrations and use all of them, paying a fraction of the
cost, right, through the waterfall enrichment. That's huge. Another thing, which is about adding
contacts, you can use third-party scrapers like Apify. You can use an actor in Apify, pulling
data to Clay and then exporting it and manipulating it the way that you want. So you can even build
custom AI scrapers. You can use Claytion to go and fetch information or data from a website,
or you can grab unstructured data like quotes from a podcast or, for example, niche attacks, stacks,
buried deep websites. So enrichment is where your list starts becoming actionable.
And now let's talk about transform, which is basically clean up and customize your data.
So now that we have enriched your data, it's time to transform it. This means getting it into the
exact shape you need for your tools or outreach, right? So you could use, for example, use AI to
summarize articles, interviews, or social bills, for example, a LinkedIn profile. You can also
generate personalized snippets. You can use Clay AI blocks to basically write a good code email
or a LinkedIn message. That's super helpful. You can also use Clay functions to basically
normalize like company names, job titles, locations. This is very helpful to dedupe
records in your CRM. For example, you have, you don't want a company name like
John Jones LLC or John Jones, blah, blah, blah. No, you want just the name, right? You don't want the
fluffy that goes after it. Okay. So this is how you use that function to normalize the names.
That's it. Now let's talk about export, which is to send where it matters, right? Finally, the export
step where all that work gets shipped to the right place, right? You can send enriched and
clean data to, for example, the CRM like HubSpot, Salesforce, Pipedrive, where basically marketing
and sales can act on it. Or you can use sequencing tools like Outreach, like Instantly, Smart Lead,
Apollo, where your leads are basically slotted into custom sequences with personalized messaging.
Or also you can use just basically a data warehouse. You can think, of course, a lot simpler,
a Google Sheet, or you can use NoFlake or BigQuery, which is perfect for large and more
enterprise scenes to centralize data across departments. So export is where your workflow
ends, but your campaign and go-to-market motion actually begins. So this is super important.
Okay, guys, let me give you one final tip. If you are ever stuck on how to start, you can use this
tip always, okay? So draw your workflow on paper first. So if you have a tablet or even a paper,
a pen and paper, sketch it out. You can draw, make the connection. So that's super helpful.
So that way you can actually see and process the information on how this workflow actually works.
You can map what is a step, what steps are defined, what is the enrich, what part of this
needs transforming, and where does it get exported. Map all of those things, map the FET framework.
You'll be surprised by how quickly this helps you just unblock complex builds.
Okay, so let's dive in the sources in Clay. So what are sources in Clay? A source is just how
you get data into a Clay table. And you can think of it as your starting point, your raw material.
In a traditional go-to-market tool like Zoomif or Apollo, you can only start by searching inside
their database, right? But in Clay, you can start anywhere. For example, you can go for
native sources like Find People, Find Companies, or Find Jobs, or external sources like Google Maps,
if you're targeting local businesses. Even web scraping like Appify, if you're trying to pull
data from websites, for example. Another example, if you're targeting brick and mortar companies in
Midwest cities, those businesses might not show up on LinkedIn. But with Clay, you can scrape
them from Google Maps and build your list. The key takeaway here is Clay gives you more doors
into the data, right? You're not limited by what a single provider can offer. Okay, so now,
once you've picked a source, you'll usually be building one of two types of lists of people
or lists of companies. Let's break those down. If you're using the Find People source, you're
starting directly at the contact level. You might search for marketing directors at a Series A SaaS
company in North America. That gives you names, roles, titles, experience, and so on. Everything
you need to start outreach. If you're starting with Find Companies, you're building a company
level list first, right? And then enriching it later to find the right decision makers.
This is especially useful if you want to do account-based research before going after
individuals. For example, you might find all Shopify Plus stores and then enrich that list with
the head of ops contacts later, right? Starting at the company level is often more flexible
for multi-threaded outreach or when doing like a custom enrichment by domain.
Now, what if you're not building a list from scratch? Clay isn't just for outbound prospecting.
You can also pull in your first party data from CRMs like Salesforce or HubSpot. Let's say you
have got a list of MQLs in your CRM, but the data is outdated. You can connect your CRM to Clay
and sync that list into a Clay table. From there, you can clean it, enrich it, push it.
Here are a few common use cases. For example, you can fill in missing fields like company size,
LinkedIn profile, or work mail. You can score leads based on job title, tech stack, or recent
hire activity. And it's like giving your CRM superpowers without needing to export spreadsheets
every week. So to wrap up, let's do a summary here. Clay offers a flexible, modular way to bring
data in, whether it's external data sources or internal ones like your CRM. You can start
building lists of people, of core companies, depending on your goal. You can use Clay
ecosystem of sources from Google Maps to your like the CRM and to go way beyond
what traditional prospecting tools offer. Most importantly, this is just the beginning.
So once your data is inside Clay, you'll be able to enrich it, transform it, and automate the entire
thing. But none of that matters if you don't get your sources right first. So in the next lesson,
we go deeper into enrichment, AI workflows, and so much more. So now let's do an exercise together.
Okay, so now that we covered exactly the FET framework and also the basics in Clay,
it's time for us to put that in practice. Today's exercise is one of the most common use cases in
Clay. It's scraping Google Maps to find local businesses. What I want you guys to do is try
it yourself. Don't go straight to the tutorial. Give yourself a chance to get to try and build
this Clay table. That way you will learn exactly how you can do that instead of just watching me
building it. Okay, what is the exercise then? We are going to find every marketing agency in Austin,
Texas using Google Maps and then find the CEO of each company. I will say that again.
Find every marketing agency in Austin, Texas using Google Maps and then find the CEO of each
of those companies. Try and do that yourself. Otherwise, here we go. So let's start the
tutorial. First thing, you need to create a Clay account. Very important. Go over to the description
and click on the link that I provided. Just go over there. That link will give you 3,000 free
credits. So try and use that. Create your Clay account and this is exactly what it's going to
look like. A brand new workspace. So nothing here and that's where we begin. So let's dive in. What
we want is marketing agencies in Austin, Texas and then I want to find the CEOs of those companies.
If this gets too fast, feel free to slow it down and re-watch it. Okay, but let's start.
Local business and then I'm going to type here Austin, Texas. Remember to adjust the radius here
just to match exactly the location that you want. So you can do, let's do like this and then you're
prompted with two types of searches. One is a business type and the other is a pretext. Business
types is just a drop-down list of businesses, of course. If your business, the one that you want
to scrape is in here. Let's see. Marketing agency, no. So marketing agency is not here. What do I do
now? I go for pretext. So pretext, you can type in exactly what you need. So that will work for
some use cases, but if you want, for example, astronauts, just go for the business types and
you're 100% sure that it would be astronauts in there. So let's do marketing agency and now let's
do, of course, you can do, always just start small. Remember, never go, remember you have limited
credit. So let's do 10 and so should be good, right? So Austin, Texas, pretext, marketing agency. Let's go.
So there you have it, guys. So Clay scraped Google Maps and found some marketing agencies.
Very important that you always want to check if they are indeed marketing agencies. So let's see
if that's the case. Let's open a few links. This one seems like it. This one seems like it as well.
This one as well. Perfect. Okay. So let's assume that all of those businesses are indeed marketing
agencies. Now what we want to do is search, find, find the CEOs of those companies. How do we go about
doing that? We only have the domains. We only have the names and the Google profile. The first thing we
need to enrich the company to find LinkedIn profiles of those, say, companies. Let's do that.
So let's do website. The company identifier is either a LinkedIn URL or a company domain.
So here we go. Website. Now this is where I'm going to teach you on another class, but let's
leave it as is. Continue to add field. Now what we need is a company LinkedIn URL. So you can pick
and choose exactly what you need in here. So let's do and let's not save. I will just run a few.
I will just run five rows just to save on credits. Let's do that. Okay. And now it's going to enrich.
Perfect. So we enrich the company with more data. So now we know employee count, LinkedIn URL for
those companies. So let's find the CEOs of those companies using the find contacts at company
enrichment, native enrichment. Remember all of those that are gray are native in clay. So they
are higher quality. Okay. So if you have the option, I always go for that. Save results in
a new table because each row is going to be one different person. Method search by search query,
because you're searching within a clay table. Very important guys. Now the filters we need to pick
exactly the LinkedIn URLs for the companies that we scrape. So you can click here, go to
the table that we just created. And then you see right here, it's mapping out in the column for
the LinkedIn URLs for those companies. Pay attention to that. Otherwise just going to
scrape random CEOs in random companies. So we don't want that. So let's do job title,
owner, partner, CISO, VP, director. Actually let's try and do just CISO. Let's see what happens.
Now let's do limit per company. Let's do five, four, and let's do 10 total. There's a bunch more
filters you can play with. Let's do that. There you go guys. So now we have the CEO and founder
of some of the companies. Remember we only enriched close to six and we found three only
CEOs. So we can add more titles in here. We're probably getting way more. So let's import that.
You can enrich the person to find, for example, more information about them, such as email,
work email, mobile phone. So in this case, let's not do that. Let's go with the free one.
There you have it guys. So we found in these businesses, these marketing agencies,
by the way, remember we did not scrape all of them. We could have scraped more than 10 results.
You could have scraped a hundred results just to save credits so you can play with your own
account as well. And then what we did next is we enriched the company to find their LinkedIn URL
because that's needed to find contacts that worked at those companies. So that's what we did.
Enriched URL, LinkedIn URL, and then find contacts at company. And there you have it.
So now we have the CEOs of a few of the companies that we actually used. Hey everyone, welcome back
to the Clay course. So in chapter two, we are diving into something that makes Clay incredibly
powerful, which is AI agents and AI formulas. And honestly, I think it's one of the most important
classes of this course. So pay attention. This is where Clay really starts to feel like magic,
but to make the most of it, you need to learn the basics of prompt engineering and how to make your,
for example, instructions for AI and how to use formulas to automate everything from
data cleanup to smart workflows. So let's get into it. At the heart of using Clay's AI effectively
is prompt engineering, which not only applicable in Clay, but you can extrapolate that to several
other, not only tools, but even your professional career as well. Don't let that technical term
intimidate you. Think of it as learning to clearly communicate your needs to an incredible,
powerful assistant. It's about crafting the right instructions so AI delivers exactly what you're
looking for. A few cases can be, for example, summarizing a LinkedIn profile in seconds or
extracting key insights about a product from a scrapped website or cleaning up
messy job titles, like transforming senior marketing manager global into a standardized
marketing manager. All of those are super useful, right, for our list building work. So the beauty
of prompt engineering lies in its flexibility. Prompt could be a single word or several detailed
paragraphs. We will learn how to guide the AI to handle these nuances with ease, okay.
So let's talk about crafting effective prompts and also the power of iteration. So what actually
makes a prompt good? In short, it's a mix of practice and continuous refinement. There's no
one-size-fits-all solution. You need to test, observe the AI's output, and then iterate on your
instructions, okay. So here's a crucial tip. Especially when you are starting out, always test
your prompts on 10 or fewer rows of data first. We will talk a lot more about that on the exercise,
but remember this. Always test your prompts on 10 or fewer rows. Always test your prompts on 10
or fewer rows. Why is that? Because you're paying for credits, guys, and each row is a credit. So
always test it first on a few rows and then scale. This is absolutely vital. Why? Because
you want to be certain of your prompt yields, your prompt is actually accurate before you scale to
hundreds or thousands of rows, okay. Imagine spending valuable credits on a prompt that
isn't quite right. By starting small, you preserve your credits and gain confidence in your approach.
Let's talk about now the official Clay prompts template, which is basically your blueprint
for success. To help you build consistent and accurate prompts, Clay provides basically an
official, a highly recommended structure. Think of this template as your reliable blueprint for
interacting with AI. It's logically divided into distinct sections, each marked with hashtags,
which clearly signal to the AI what information to find within each part. The best way is to
actually use AI to create a prompt. I will show you where that is in Clay's platform in today's
exercise, but remember, the structure is always the same. So I will share the structure, this
entire blueprint, as a doc, as a Google Doc, and you can just copy and use it. If you don't, of
course, if you don't want to use the AI function in Clay, but just use it. This is not only valuable
to, as I said, to Clay, but your career if you are using other AI tools as well. Good? All right,
next up, let's discuss how to pick the right AI model for your prompts. So this is a common
question for new users, but right simply means finding the model that offers the best blend
of accuracy and cost efficiency. If you are ever unsure, you can click, I think that was,
I'm sorry, a button called Compare Models. Within the prompt interface, you see a quick
table summarizing costs and costing credits, of course, and key characteristics of each model.
Our recommended strategy is this. You can start with the most capable one, by the way,
or I say ours, but this is Clay and myself. So start with the most capable,
often the most expensive model. Run your prompt on only 10 rows of data.
Manually check the accuracy of the results. If the results are accurate,
you can clone that column and then select the next lowest cost model. Compare the results
between the two columns and ask yourself, are they still accurate? If so, that's great. You
might have found a more budget-friendly option that performs just as well. This iterative approach
helps you pinpoint the optional model, ensuring that you get excellent results without overspending
Now, let's shift our focus to AI formulas. I truly, honestly, underutilized an incredible,
powerful feature in Clay. And the best part, guys, they don't cost you any credits. This is huge.
Imagine being able to generate custom action or logic directly within your Clay table
without writing a single line of code. That's precisely what AI formulas empower you to do.
We can categorize these formulas into three key areas, each offering, of course, significant
advantages. So first one, conditional runs. This is a major credit saver. For example, let's say
you only need to enrich data for companies with over 100 employees. You can write an AI formula
that instructs Clay to only execute if the company size column is greater than 100. If the
condition, of course, is not met, Clay simply skips that row. This kind of conditional logic can save
you hundreds, even thousands of credits on large data sets. So remember this. Data formatting.
Formulas offer a fantastic credit-free way to clean and standardize your data.
Whether you need to merge two separate names into columns into a full name or standardized
phone numbers formats, AI formulas can also handle those. So make sure to use it.
And also conditional outputs. This allows you to dynamically manipulate the output of a column
based on conditions in other cells. Think of it as building smart, if then, logic directly into
your table. So common use cases. If the email address contains at gmail output not workmail.
If the company size is less than 50, mark delete as small business. So these formulas enable you
to add basic intelligent automated classifications and also transform your data without any
additional AI costs. Now guys, to wrap up. Chapter 2 has equipped you with some truly essential
skills for leveraging Clay AI. We have explored the nuances of prompt engineering, emphasizing
importance of iterative testing. We also guided you through Clay's official prompt template. Make
sure to grab the Google Doc file. Also discuss how to intelligently choose the right AI model
and introduce you to maybe one of the most powerful and also credit-free features in Clay,
which is AI formulas. So keep practicing, keep experimenting, and remember that mastering these
tools will significantly enhance your productivity. So let's do now the exercise together.
Welcome back. So now in this exercise we are going to take a messy list of contacts from LinkedIn
and clean up their job titles and then score them as potential leads. How? Without writing
a single line of code. We'll cover a bit of prompt engineering as well in this exercise,
choosing like the right AI model and using credit-free AI formulas. Let's dive in and also
remember guys, try and do the exercise yourself first and then you go and watch the tutorial. I
will name the exercise in a few minutes and try and do it yourself. That way you learn exactly how
you can do it and that way it's a better learning experience. Okay, sit tight and let's dive in.
The exercise that I want to do with you guys is use an AI agent to visit
the website that we scraped from the Google Businesses profile and
fetch information about their service. In this particular case, I want to know if they sell
brand services, right? If they sell brand services. And then if yes, I'm going to score that lead
with a point just to create a lead score system, okay? So again, the exercise is use AI agent to
visit the website that we scraped, find if they sell brand services, if yes, you score them a point.
Okay, try and do that yourself and let's hop right into the tutorial. First thing that we need to do
guys is use an AI formula, which in this case is already native in Play, which is to
normalize the company domain. Why do we want to do that, Guy? Because, I'm sorry, if you see right here
the website is very messy. By the way, this is the exact table that we built in chapter one,
exact one, okay? The exact same businesses. The websites are a bit messy, so they have UTM, they have
WW, I don't like that. I wanted a clean column. So let's do that. Normalize. Actually, I think
to just normalize. And then normalize the domain. Now we have to map the exact name of the column.
It's not website2, it's just website. So let's pick this normalization type. I just
want the domain isolated and now let's run all columns. All right, guys, much better. You see here
much cleaner. So this is really good. So now let's use AI to visit the website and fetch if they sell
brand services. How do we do that? Let's use Clasient. Use AI, right? Let's go to configure,
okay? What is the use case that we want? We want the agent to have internet access because it's
going to visit the website. So we definitely need Clasient. So let's do that. A model. Let's use
Argon first and then we can play with it. Usually it's pretty good and also it automatically adjusts.
So you can do Argon, you can do Neon, it's up to you. So let's try Argon. Now we click generate
because I want to view the prompt. Now remember about the prompt engineering tips. So first,
be specific. We need to say I want you to visit and then you do slash website right here, okay?
Actually, what's the name of the, it's actually normalized, right? It's normalized domain and
then you're going to use normalized URL, okay? This is where we have the websites. By the way,
why we are doing slash and pick a column? Because otherwise it's just not going to know where you're
getting the data. So I want you to visit slash normalized URL and go to the service page or
about us page or in the home page and check if they're, if they sell brand services. If yes,
output. If yes, output yes. Otherwise, no, okay? We need to be very specific here, guys. I want
to visit this website and this particular page and this page or home page and check if they sell
brand services, okay? So now it's going to, this is the meta prompt and now it's going to generate
the prompt for the agent. All right, so we have it. There we have it. So we have the prompt in
here. You just need to read it and see if they make sense. So let's take a look. Since yes,
so let's see. Let's do run. I will just run a few rows. I don't want to run 10 rows.
So let's do that. Let's run one, two, three, four. Okay, so quite a few of them actually offer
brand services. The good thing about Clay, you can actually see the reasoning. If you click on the
cell, you see exactly what the agent has done. So you can read it. It's a creative agency. They offer
services such as brand campaigns, et cetera, and these steps taken. So you can see it visited the
work page, the work tab and got the information. All right? So now we have at least some
categorization, some classification. So let's do an AI formula. I want to do, and now we are going
to use a meta prompt again. If actually what's the name of the column? The brand services offered.
Let's rename it to brand services. Okay? Now let's add a column formula and then we can type in
if slash, always remember to add the column, if slash brand services is
categorized as tier one. If brand services is no, categorize it as tier two. Okay?
There you go. There you go, guys. So now we have a categorization that is very, not, it's very
simple, honestly, but remember that you can build whatever you want with Clayton. So this helps you
a ton in terms of categorizing companies, target accounts and so on. So always remember to do that.
Now let me give you one last tip. Remember when I said where to use the AI formula to save credits?
This is where you can use it. If imagine we, we didn't have some domains in here. In case, for
example, we scraped the business profile, but they didn't have a website. We actually only want to run
where, only run where website is not empty. Okay? By the way, in this case, it would work anyways,
because the website is one of our inputs. But one of the things that you can try as well is
only run where, let's say, description says is, for example, is mentions brand services.
So you can play with this in a way that it will save you some credits. So that's it. That's one
final tip for you guys. But this is pretty much of the exercise. Play with it. You can, this here
is one of the most cost-saving opportunities we have in Clay, as well as the AI formula. So
there you have it, and see you in the next. Hey everyone, and welcome back to the Clay course. In
our last session, we laid out the basically the groundwork for the FET framework. Today,
we are diving into one of the most powerful time-saving and cost-effective features in the
entire Clay platform. A feature that was basically, I think, the talk of the town long
before AI agents came onto the scene. We are talking about waterfalls. By the end of this
session, you know how to use different types of waterfalls to find almost any data point that you
can imagine. How to auto-conserve your credits while doing it, and even how to build your very
own custom waterfall from scratch. Let's get started. So, what exactly is a waterfall?
Imagine you need to find someone's email address. You could use a data provider like Hunter.io,
but what if Hunter comes up empty? You then have to try another provider, let's say,
Drop Contact. And if that fails, maybe another. A waterfall in Clay automates this entire process.
It lets you chain together multiple data providers, one after the other. It tries the first
one. If it finds data, it stops. If it fails, it automatically moves to the second, and then the
third, and so on, until it finds what you're looking for. In the best part, you only pay for
the one provider that actually finds the data. It's a sequential, cost-effective, and incredible,
powerful way to enrich your data. Let's take a trip back in time. So, what did life look like
without that? Imagine your task is to find verified email addresses for a list of a thousand leads.
The process was painfully main. First, you would export your list and upload it to an email provider,
let's say, Provider A. You would get your results, but you're not done. Second, you have to take that
list of new emails and run it through a separate validation service, like Zero Bounce, to see which
ones are valid or catch-all or invalid. And third, for all the leads that were still missing an email
or had an invalid one, you would have to start the whole process over again. You take that segment,
upload it to Provider B, get their results, and run those through the validation service, and so on.
This wasn't just for emails. The same tedious, multi-step, manual process applied to finding
phone numbers, company revenues, tax, any data point that you can think of. It was slow and
inefficient and expensive as well, because remember, you're paying each of those providers separate.
Now, let's see how Clay transforms this. With a waterfall, you can find a contact email,
validate it, and have multiple backup providers ready to go, all within a single column and with
just one click. See the difference? Instead of a tangled web of exports and imports, you have one
clean, linear flow, which is much better. So Clay tries the first provider. If it fails, moves to
the next one. You can even insert a validation step right inside a waterfall between each provider,
ensuring you only get verified data. So this isn't just for emails. This also applies to
everything. You need to find a company's headcount, set up a waterfall with providers like Apollo,
Clearbit, People Data Labs. For example, if you need to find their tech stack, waterfall it.
So this is the new and streamlined way of enriching data. All right, the theory is great, but let's
get our hands dirty. It's time to build our first waterfall. The key thing to remember is that a
waterfall is just a series of data enrichments linked together. So Clay has pre-built templates
for the most common ones, which makes getting started incredibly easy. So let's walk through
an example. The pre-built templates are fantastic, but what if you want to create your own? Maybe
you have a favorite provider or you want to use AI to scrape a specific data point. So Clay makes
this easy, just a waterfall option. From there, you can add any number of steps, chaining together
different data providers, API calls, or even AI actions to create a completely custom enrichment
flow tailored to your exact needs. And that's a wrap on waterfalls. It was, I guess, a shorter lesson
today, but the concept is a cornerstone of working efficiently in Clay. So to recap, today we learned
how waterfalls can save you hours of manual work and credits, understood the different types of
data we can find using waterfalls, and discovered how to build our own custom waterfalls. Now it's
time to put what you have learned into practice.
