Clay helps GTM teams find the right prospects, fill in the data they’re missing, automate research, and get that data where it needs to go. This guide walks through Clay enrichment, waterfalls, Claygent, prospecting, and building workflows from start to finish.
Clay can do a lot…which is exactly why figuring out where to start can be tricky.
The key is to stop thinking about Clay as a long list of features and start thinking about what you’re trying to build.
That’s where the FETE framework comes in. It stands for “Find, Enrich, Transform, Export” and gives you a simple way to take a GTM workflow from raw data to something your team can actually use.
In this guide, we’ll use FETE to break down Clay for GTM automation step by step, including how to find the right prospects, enrich and research them, clean up your data, and get it into the systems where your team works.
Clay is a platform for building and automating GTM workflows. Instead of jumping between different tools to find prospects, fill in missing data, research accounts, clean records, and prepare them for outreach, you can connect those steps in one place.
That’s what makes Clay different from a traditional database. Instead of searching for names and contact information, you’re building a workflow around that data, whether the goal is prospecting, enrichment, lead scoring, segmentation, personalization, or getting cleaner data into your CRM.
A Clay workflow can handle a lot of the manual work that happens between identifying a potential buyer and actually doing something with that information.
You can use it to:
The exact workflow will look different depending on what you’re trying to accomplish, but the building blocks are largely the same.
The upside of Clay is its flexibility. The downside is what Clay calls the “blank canvas problem.” When there are so many things you could build, knowing what to build first can be the hardest part.
You don’t need to learn every Clay feature before you get started. It’s much easier to learn the basic workflow first, then add the tools and enrichments you need as you go. That’s where FETE comes in next.
The easiest way to make sense of Clay is to follow a repeatable process. Clay calls this FETE: Find, Enrich, Transform, Export. Each step builds on the one before it, taking you from a starting list to clean, useful data that’s ready to put to work.
Every workflow starts with the people or companies you want to work with. You might find companies that match your ICP, people with specific job titles, open jobs, or local businesses through Google Maps. You can also start with data you already have and bring it into Clay.
The goal at this stage is to build the right starting dataset.
Once you have that list, enrichment fills in what you don’t know. Depending on the workflow, that could mean finding emails and phone numbers or adding company details like headcount, funding, industry, technology, and hiring activity.
This is also where you can go beyond standard Clay data enrichment and use AI-powered research to find more specific information about a company or person.
Raw data isn’t always ready to use as-is. The Transform stage cleans and reshapes it so it works for whatever comes next.
That could mean normalizing company names, standardizing phone numbers and locations, extracting information from existing fields, calculating values, or formatting data consistently. The goal is to turn everything you’ve collected into clean, usable data.
Finally, the data needs somewhere to go. Export moves the finished records out of Clay and into the tools where your team actually works.
That could be HubSpot, Salesforce, Google Sheets, a sequencing tool, a CSV, or another GTM system. Once the data gets there, it can actually be used for prospecting, outreach, reporting, or other GTM work.
FETE gives you the steps of a Clay workflow, while the Jigsaw Framework helps you figure out what data you actually need along the way.
The idea is to approach Clay data enrichment like putting together a puzzle. You don’t start by filling in random pieces. You start with the information that makes everything else easier to find, then build from there.
The corner pieces are the basic identifiers Clay can use to connect a record to additional data.
For companies, start with:
These give you a strong foundation for finding and enriching additional company information.
For people, start with:
From there, you can build out a more complete picture of the person and their role.
Once the corner pieces are in place, add the basic information that gives each record context.
This could include:
For contacts, that could include:
At this point, you have the basic shape of the account or person. Now you can get more specific about the information your workflow actually needs.
The middle of the puzzle is where Clay data enrichment can become much more specific to your use case. You might add funding data, hiring activity, technology, website information, AI research, or other attributes that help you better understand and segment your prospects.
You don’t need every available data point. The goal is to fill in the pieces that help you make a better GTM decision.
FETE and Jigsaw solve two different parts of the same problem. FETE gives you the workflow, while Jigsaw helps you decide how to build the data within it.
This means you can:
Together, the two frameworks give you a more structured way to build in Clay instead of adding data and enrichments without a clear plan.
Once you understand FETE and Jigsaw, it becomes easier to see how those frameworks turn into actual GTM workflows. Clay GTM automation can support everything from outbound prospecting to local business research, but the same basic process applies: start with the right audience, add the data you need, and turn it into something your team can act on.
See how Clay and HubSpot turn enrichment, intent signals, and account scoring into an assisted prospecting workflow 👇
A B2B outbound workflow might start by finding companies that match your ICP. From there, you can enrich those accounts with information like company size, industry, funding, technology, or other criteria that help you narrow the list.
Once you know which accounts you want to pursue, you can find the right people at those companies, enrich their contact information, and add research that makes your outreach more relevant. The finished records can then be sent to your CRM or sequencing tool for activation.
The workflow looks something like this:
See how reps turn Clay signals into a repeatable HubSpot prospecting play across email, calls, LinkedIn, and video 👇
Without a workflow like this, much of the same work has to happen manually. Someone has to build the list, research each company, find the right contacts, track down contact information, and prepare the data for outreach. It’s part of why sales reps spend 60% of their time on non-selling activities.
Clay lets you connect and automate those steps, so your team can spend less time on the work around selling and more time actually selling.
Clay GTM automation isn't limited to companies that are easy to find in traditional B2B databases. You can also use Google Maps as a starting point to find local businesses such as dentists, restaurants, auto repair shops, contractors, and other service businesses.
From there, you can enrich those listings with additional information to determine which businesses fit your target market and give your team more context for outreach.
Sometimes the information you need isn't sitting neatly in a database. Clay can combine the company and people data you've already collected with websites, social profiles, Claygent, and AI to research more specific questions.
That could mean understanding what a company offers, identifying something relevant on its website, or gathering context that helps you tailor your outreach. And that context matters when 73% of B2B buyers actively avoid sellers who send irrelevant messages. The goal is to uncover information that makes the message more relevant to the prospect.
See how Clay can automatically research inbound leads and give sales the context they need before a call 👇
The same building blocks can be applied to plenty of other GTM workflows. Clay can be used for webinar attendee enrichment, website research, competitive intelligence, account-based marketing, lead scoring, and inbound qualification..
Clay gives you a lot of room to test new ideas without committing to a massive workflow from the start. Instead of trying to get everything perfect upfront, start small, see what the results tell you, and improve from there.
See how Clay can turn a large list into a focused audience before you start enriching 👇
You don’t need the perfect audience, enrichment strategy, or message before you start building. Clay makes it easier to test those assumptions with real data first.
Maybe you think a certain segment is a better fit, a particular signal shows stronger intent, or one message will resonate more with a specific audience. Build a small test around that idea, look at what happens, and use what you learn to shape the next version.
Set aside around 10–15% of your monthly Clay credits for R&D. This gives you room to test new ideas without taking credits away from workflows you already know are working.
You can use that budget to experiment with:
Long story, short: know what you’re trying to learn before you start spending credits on it.
One of the easiest ways to waste Clay credits is to run a new workflow across a huge list before you know whether it works.
Here’s a better approach:
Start with five records. Make sure your inputs are mapped correctly, the workflow runs as expected, and the output is actually useful.
Once the workflow works, move to 50 records. A larger sample helps you see whether the results are consistently good or whether those first five were just good examples.
If the workflow is still producing the results you want, start scaling to 500 records and beyond.
When something works, document it.
Keep track of the audience, signals, enrichments, messaging, and workflow that produced the result. Over time, those successful experiments become a library of GTM plays your team can reuse instead of starting from scratch every time.
Testing before you scale helps prevent wasted credits, but there are also ways to make the workflows themselves more efficient.
Not every step requires a paid enrichment or AI call. If you already have the information you need, first see whether Clay can clean, calculate, transform, or route it without spending additional credits.
Here’s what that looks like in practice when cleaning and qualifying trigger data before enrichment 👇
AI Formulas are useful when you already have the underlying data and need to do something predictable with it.
For example, you can use them to:
AI Formulas don’t consume Clay credits, so they’re a good fit for tasks that don’t require finding new information.
A prompt that works poorly across five rows can get expensive when you run it across thousands.
Use Clay’s Metaprompter to tighten your prompt before scaling an AI workflow. Make the instructions clear, check that the output is what you actually need, and then put more rows through it.
Not every record needs every enrichment.
Conditional runs let you trigger an action only when certain criteria are met. You might run another enrichment only when a company qualifies for the next step, for example, or only when the information you need is still missing.
That keeps you from spending credits on records that don’t need the additional work.
Use the Sandbox to test prompts, enrichments, and workflow ideas before putting them into a larger production workflow.
It gives you a place to see what works, adjust what doesn’t, and get the output right before you start using more credits.
You can also bring your own API keys for supported providers or AI models. Depending on the workflow, this can give you another way to manage how and where you’re spending Clay credits.
If you need Clay to find new information or perform outside research, spending credits may make sense. If the information is already there and simply needs to be cleaned, calculated, formatted, or routed, use the simplest option that gets the job done.
The first step in FETE is Find, and for a typical B2B workflow, that often starts with building a list of companies that match your ICP.
Clay’s Find Companies tool lets you narrow that list using criteria like industry, company category, headcount, location, keywords, and founding date. Think of this as your starting dataset. You can always enrich those companies later when you need more specific or up-to-date information.
Start by creating a workbook. If you’re used to spreadsheets, the setup is pretty familiar as a workbook is similar to a spreadsheet, while individual tables work like sheets within it.
From there, open Find Companies and start defining your ICP. You can combine filters like industry, company size, geography, and keywords to narrow the results to the types of accounts you actually want to target.
Before importing anything, preview the results. Look at the companies Clay is returning and check details like:
If the list looks too broad, go back and adjust your filters. It’s much easier to tighten your audience here than to clean up thousands of irrelevant records later.
Once the results look right, import the companies into your table. You don’t need to start adding every possible enrichment yet. First get the audience right, then decide what additional data the workflow actually needs.
Bigger isn’t always better when you’re building a company list in Clay. A focused list of roughly 1,000–3,000 companies gives you a more manageable audience for testing than one massive list with loosely related accounts.
More importantly, build each list around an idea you can test. For example, you might build a list around recently funded SaaS companies, companies that are hiring aggressively, or bootstrapped companies with technical founders.
Your inclusion criteria tell Clay what you want, but exclusions can be just as important for removing what you don’t.
If your search keeps pulling in agencies, consultants, or another type of company that doesn’t fit your ICP, use negative filters to remove them. The cleaner the starting list is, the less time and credits you’ll spend dealing with irrelevant companies later.
You don’t need to enrich everything yet, but you should think one step ahead.
Make sure the company records you import include the identifiers you’ll need later, especially the company domain and LinkedIn company URL. Those are the “corner pieces” from the Jigsaw framework and give Clay a stronger foundation for the enrichment steps that come next.
Once you have the right companies, the next step is finding the right people within them. Instead of building a separate contact list from scratch, Clay lets you use your existing company table as the starting point and search for people who match your target persona.
Before you begin, you’ll want three things in place:
Start by translating your target persona into filters Clay can use. Depending on who you’re looking for, you can narrow the search using criteria like:
You can combine these filters to get as specific as your workflow requires. The goal is to find the people who actually make sense for the companies you’ve already identified.
From your company table, use Find People at These Companies and map the company identifier that connects your search back to those accounts.
Then add your persona filters and preview the results before importing them. Check whether Clay is returning the roles and people you expected. If the list is too broad or you’re seeing irrelevant contacts, adjust the criteria and preview it again.
Once you’re happy with the results, import the contacts into a people table. Now you have the two sides of your prospecting workflow connected: the companies you want to target and the people you may want to reach within them.
Your first people search doesn’t have to be your last. You can go back and adjust the criteria as you learn more about who you want to reach.
For example, you might start with a narrow set of senior titles and later expand into additional roles or functions. That lets you build on the company list you already have instead of rebuilding the entire workflow every time your targeting changes.
Jobs can be useful in Clay in two different ways. You can start with open roles and use them to find companies, or start with companies you already care about and use their hiring activity as another layer of data.
Which approach makes sense depends on what you’re trying to learn and where you want the workflow to start.
The biggest difference is your starting point.
When jobs are the source, you start by looking for specific open roles and then work backward to the companies hiring for them.
This can be useful for recruiters and job seekers, but also for sales and BDR teams looking for companies with a specific hiring need.
When jobs are an enrichment, you already have a list of companies that fit your ICP. Instead of using hiring activity to find those companies, you add it to learn more about them.
That information can help with segmentation, prioritization, scoring, and understanding what’s happening inside an account.
Start by defining the types of jobs you want to find. You can search broadly across companies or narrow the search to a specific group of companies.
From there, add the job titles you’re interested in and use criteria like location and job type to refine the results. Preview the number of jobs Clay finds before importing them. If the list is too broad, adjust your search until you’re getting the kinds of roles you actually want.
Once the results look right, import the jobs into your table and continue building the workflow from there.
The process works a little differently when you already have your target accounts.
Start with your ICP-fit company list and add job data as an enrichment. You can search for specific roles to see which companies are hiring for the people or skills that matter to your use case.
Depending on what you’re looking for, you can use that data to understand things like:
The value is knowing what that hiring activity can tell you about the company.
For example, hiring patterns can give you more context around where a company is growing and what it may be prioritizing. That turns job data into another signal you can use to segment and prioritize accounts.
Job searches usually get better as you refine them. A few practices can help you get more useful results from the start.
Job search isn’t fully semantic, so one title may not capture every relevant opening. Include different variations of the roles you care about rather than relying on a single keyword.
You don’t have to get every filter right on the first search. Start with your main criteria, preview what comes back, and then narrow the list using job titles, location, job type, and other available filters.
Job data can come from different underlying providers, and those providers can have their own search and API limitations. Keep that in mind if the results are different from what you expected.
Treat the first search as a starting point. If you’re seeing irrelevant jobs or missing roles you expected to find, adjust your keywords and filters and run it again.
The goal is the same as the other Find workflows: get the starting dataset right before spending time and credits enriching it further.
Not every prospect is easy to find in a traditional B2B database. That’s especially true for local businesses like dentists, restaurants, auto repair shops, and other service-based businesses.
Clay lets you use Google Maps as a source, giving you another way to build prospect lists based on the types of businesses you want to reach and where they’re located.
Google Maps is particularly useful when your target market includes businesses with a strong local presence but a limited footprint in traditional company databases.
Instead of starting with a company database and filtering down, you can search for businesses in a specific area and bring those results directly into Clay. From there, the Google Maps listing becomes the starting point for further research and enrichment.
Start by choosing Google Maps as your source in Clay. From there, define where you want to search. You can use a city, neighborhood, exact address, or radius depending on how targeted you want the list to be.
You’ll also need to decide how you want to search:
When there’s a relevant Business Type available, it can give you a more structured way to search than relying entirely on free text.
Once your criteria are set, run the search. Keep in mind that Google Maps results can continue populating after the search begins, so give Clay time to finish pulling in the businesses before deciding whether you need to change your criteria.
Finding the businesses is only the beginning. A Google Maps listing gives you a starting record, but Clay lead enrichment can add the context you need to decide whether that business is actually worth targeting.
For example, you can look at ratings and review counts to help evaluate businesses, then add other information such as:
You can also use AI to analyze Google Maps reviews. Instead of reading through reviews one by one, you can use that information to better understand the business, what it offers, and what customers are saying about it.
The exact data you add should depend on the audience you’re building. A dentist prospecting workflow may need different information than one targeting restaurants or auto repair shops.
Once those details are added, a basic Google Maps result becomes a much more useful prospect record that can be qualified and eventually used for outreach.
A few choices at the Find stage can make the rest of the workflow much easier.
Define your search area carefully. If you’re targeting multiple locations, separate searches can help you keep each audience focused instead of trying to cover too much geography at once.
When Google Maps has a Business Type that matches the businesses you want, use it to create a more consistent search. Free-text search still has its place, especially when the category you need isn’t available.
Don’t assume the first result count you see is the final one. Give the search time to finish populating before changing your filters or running another search.
Before importing a large number of businesses, think about what you’ll need to know about them next. If ratings, reviews, contact information, services, or another attribute will determine whether a business is worth targeting, plan for that before you scale the list.
That keeps the workflow moving naturally from Find into the next stage of FETE: Enrich.
Finding gives you a starting list, but enrichment fills in what you need to know about the companies and people on that list.
The goal is to have the right data to qualify, prioritize, segment, and eventually act on a record.
Clay data enrichment is the process of adding new information to records you already have. Depending on the workflow, that could include company information, contact details, technology data, hiring activity, or more specialized research.
For example, a basic company record could become:
The information you add should depend on what you’re trying to accomplish. An outbound campaign may need different data than an account-scoring or market-research workflow.
Clay lead enrichment applies the same idea to the people or prospects you may want to reach.
You might start with a name, LinkedIn profile, and company, then add information such as a work email, phone number, job title, work history, or other context that helps you qualify the person or prepare for outreach. That work matters when sales reps spend 27.3% of their time dealing with inaccurate B2B contact data, adding up to roughly 546 hours per year for a full-time inside sales rep.
Again, more isn’t necessarily better. If your outreach is entirely email-based, for example, there may be little reason to spend credits finding a phone number you won’t use.
Basic firmographic data like company size, industry, and location can be useful, but it usually isn’t where the biggest advantage comes from. Plenty of teams can access that information.
The more interesting opportunities come from finding data that helps you identify something specific about your ICP that others may be missing.
Instead of targeting companies based only on size and industry, you might combine several pieces of information to identify companies that are hiring for a particular role, use a certain technology, and meet another condition relevant to your offer.
One of the easiest mistakes to make in Clay is adding data simply because you can. A table with dozens of enrichment columns may look impressive, but those fields aren’t doing much if they never affect a decision.
Before adding an enrichment, ask what you’re going to do with the result. Will it change who you target? How you prioritize the account? Which segment it belongs to? What you say in your outreach?
If the answer is no, you may not need it.
The same applies to your overall workflow. Know whether you’re enriching data for outbound, qualification, scoring, segmentation, market research, or account intelligence before you start adding providers and spending credits.
Linked tables can also make Clay enrichment more efficient by letting company and people data work together.
Say you find 10 people who all work at the same company. If you need a company-level attribute, you don’t need to enrich that same information separately for all 10 people.
Instead, enrich the company once in your company table and pass that information to the related people records. That keeps the workflow cleaner, reduces duplicate data, and avoids paying repeatedly for the same company-level enrichment.
Once you have a list of companies, company enrichment lets you add the information you need to better understand and qualify them.
Before you start, make sure you have strong company identifiers. A company domain is especially useful, and a LinkedIn company URL can provide another reliable corner piece. You should also know what you’re trying to learn before choosing your enrichments.
One of the main ways Clay helps you find that additional information is through waterfalls.
Clay waterfall enrichment checks multiple data providers in sequence to find a piece of information.
Instead of relying on a single provider and accepting a missing result when that provider doesn’t have the data, Clay can move to the next provider in the waterfall.
A simple Clay waterfall might look like:
Provider A → no result
Provider B → no result
Provider C → result found
Waterfall stops
Different providers have different strengths and coverage, so using them together can increase the chances of finding the information you need.
The biggest benefit of a Clay waterfall is coverage. Single-source data providers typically deliver only 50%–70% coverage, while waterfall enrichment can increase that to roughly 85%–95% by checking multiple providers in sequence.
That lets you take advantage of each provider’s strengths without manually checking them one by one. It can also help control unnecessary spending because the waterfall stops once the requested information is found rather than continuing through every provider.
Provider order matters, too. Depending on your workflow, you may want to balance cost, coverage, and the quality of the results when deciding which providers should run first.
Start by choosing the information you want to find. That could be something like a company’s funding stage or another company attribute required by your workflow.
Clay will show the providers available for that enrichment and the order in which they’ll run. From there, map the required company identifier, such as the company domain, so the providers have enough information to search for the right record.
Don’t immediately run the waterfall across your entire table. Test it on around 10 rows first.
Look at whether the data is being found, whether it’s accurate enough for your use case, and what the waterfall is costing you. If the results look right, you can run it across more of your list.
The answer depends on what you’re trying to accomplish. There’s no universal list of fields every company needs.
Firmographic information helps you understand the basic characteristics of a company. Depending on the enrichment available, that could include information such as:
These fields can be useful for determining whether a company fits your ICP or belongs in a particular segment.
See the different ways to find funding data in Clay and how to choose the right one for your use case 👇
Technographic enrichment helps you understand the technology a company uses.
If your offer depends on a company using a particular platform or category of software, that information can become a useful qualification or segmentation criterion.
Hiring activity can add another layer of context. Instead of simply knowing what a company looks like today, you can look at what roles it’s hiring for and use that information to better understand where it may be growing or investing.
The important part is connecting the enrichment to an actual GTM decision rather than collecting the field just because it’s available.
Individual data points become more useful when you start combining them.
Company size alone may tell you whether an account fits your basic ICP. Add technology, hiring activity, funding, or another relevant signal, and you can start creating much more specific segments.
You can also adjust the order of providers in a waterfall when needed and use conditional runs so an enrichment only happens when a record actually requires it.
The goal is to progressively add the information that makes the company more actionable without enriching every possible field on every possible account.
Company enrichment tells you more about the account, while people enrichment helps you understand and reach the individuals within it.
Ideally, you’ll already have strong identifiers such as the person’s full name and LinkedIn profile, along with the company they work for. From there, decide what information your actual outreach strategy requires.
For many outbound workflows, one of the most important pieces is a valid work email.
Clay can use a waterfall for work emails just like it can for company data. Instead of depending on one email provider, it can move through multiple providers until it finds an address.
But finding an email and finding a usable email aren’t exactly the same thing. Validation helps determine whether the address is one you actually want to use for outreach.
Catch-all domains add another consideration. They can make it harder to determine whether an individual mailbox is valid, so your settings may depend on how you want to balance additional email coverage against a more conservative approach to deliverability.
Start by adding the work email enrichment to your people table and reviewing the providers included in the waterfall.
Then map the inputs Clay needs to identify the person. Depending on the provider, those can include:
Configure your validation settings based on the level of confidence you want, then test the enrichment on a small number of records.
Again, 10 rows is a good starting point. Look at how many emails are being found, how they’re being validated, and whether the results make sense before running the waterfall across the entire table.
If you need to dig deeper, you can expose additional validation information to better understand what happened with a particular result.
Email is only one type of Clay lead enrichment. Depending on how you plan to qualify or engage a person, you can add other information such as:
Which fields matter depends on the workflow.
If you’re building a technical prospecting campaign, GitHub or work history may be useful context. If your team uses multiple outreach channels, phone information may matter more. If none of that changes how you qualify or contact the person, you probably don’t need to enrich it.
People enrichment can also go beyond filling in one contact at a time.
For an important account, you may want to find several relevant people instead of relying on a single contact. That gives you a better picture of the people involved at the company and gives your team more options for reaching the account.
The same principle from the rest of Clay enrichment applies here: start with the data you already have, add what your workflow actually needs, and avoid collecting information without a clear reason for using it.
Most of the enrichment we’ve covered so far deals with data that already exists in a structured format like company size, funding, work emails, and job history.
But some of the most useful GTM data isn’t that easy to find. You may need to know whether a dentist offers Invisalign, a company has a free trial, or its website mentions a specific technology.
Claygent is Clay’s AI research agent for finding this kind of information on the web. You give it a specific research task, provide the context it needs, and run that research across the companies or people in your table.
This is useful for what Clay calls “last-mile data,” or information that matters to your specific use case but probably won’t exist as a standard field in a traditional database.
Claygent still needs a solid starting point, so you need to give it enough context to identify the company or person you want it to research.
For company research, that’ll usually include:
For people research, you might use:
This goes back to the Jigsaw framework from earlier. Give Claygent strong identifiers first, then use it to find the more specific pieces.
Start with a specific research task.
If you have a list of companies and need to identify which ones offer a free trial, you would add Claygent to your table, write a prompt explaining what you want it to find, and map in the company website or other information it should use for the research.
Next, choose the model that’ll handle the task. There’s no reason to use a more expensive model if a cheaper one can reliably produce the result you need.
Test the workflow on around 10 rows before going any further. Read the results yourself and check whether Claygent understood the task, returned consistent answers, and gave you data you can use later in the workflow.
If the results are off, adjust the prompt and run another test. For research where you need more visibility into how Claygent arrived at an answer, you can also review its reasoning or confidence.
Once the test results are consistent, you can start running the research across a larger list.
Claygent becomes useful when your targeting depends on information that isn’t sitting in a normal database field.
You can use it to find out:
That last use case gives you much more freedom in how you define an audience. A characteristic you care about may not exist as a filter in a traditional data provider, but Claygent can research it and turn the result into something you can use for targeting, qualification, or segmentation.
Claygent also doesn’t need to handle every enrichment. If a standard provider can reliably give you the information, use it. Claygent is better suited for information that actually requires research.
Keep the prompt clear and specific. Claygent should know exactly what it’s looking for and what you want returned.
A few things help:
And keep the same testing approach you’ve used throughout the workflow. Run a small sample, read the results, refine the prompt, and scale once you know it works.
By this point in the Enrich stage, you’ve already built out the standard company and people data you need. Claygent lets you fill in the more specific information that’s harder to find and more relevant to how you actually sell.
By the time you reach Transform, you’ve found the right records and enriched them with the information you need. Now that data needs to be cleaned up before it goes anywhere else.
Enrichment pulls in the data. Transformation gets it ready to use.
Without that step, small inconsistencies can follow your data into your CRM, campaigns, reporting, and segmentation. Company names may include unnecessary legal suffixes, phone numbers may use different formats, or locations may be written several different ways.
Transform is where you clean that up before it causes problems downstream.
Clean data makes everything that follows easier, especially when only 35% of sales professionals trust the accuracy of their sales data.
Take company names. A record might come back as “Acme Inc.” That may be technically correct, but it sounds awkward if you drop it directly into an email.
Normalize the name to “Acme,” it immediately sounds more natural.
The same idea applies across your data. Consistent formatting makes it easier to build segments, automate workflows, personalize outreach, and keep your CRM clean.
Clay has built-in normalization tools for common cleanup tasks. These are deterministic transformations, meaning they follow predictable rules using information that’s already in your table rather than reaching out to another data provider.
Because of that, Clay’s native normalization tools don’t consume credits.
You can find them under the Normalize options in the enrichment panel and use them to handle many of the cleanup tasks that come up repeatedly in GTM data.
Extra spaces, tabs, and line breaks can create problems when data moves between systems or gets inserted into automated messaging.
Normalizing whitespace cleans those inconsistencies so text follows the same format across your records.
Phone numbers can come back in different formats depending on where the data originated.
Normalization gives those numbers a consistent structure by handling things like parentheses and dashes, making the data easier to use across systems and calling tools.
Location data can also be inconsistent, which becomes a problem when you want to segment or route records based on geography.
Normalizing locations creates a more consistent way of representing that information across your dataset.
These native tools should be one of your first stops when the task is straightforward cleanup. If the information is already there and you simply need to standardize it, there’s no reason to spend credits finding it again.
Native normalization handles common cleanup tasks, but sometimes you need to do something more specific with the data already in your table.
AI Formulas let you perform structured transformations without having to build complicated formulas or scripts yourself.
They’re particularly useful when you know exactly what you want the finished result to look like and need to apply the same logic across a large number of rows.
See how Clay formulas can clean a messy company list and prepare it for prospecting without wasting credits 👇
AI Formulas use AI to create the logic needed to manipulate, calculate, extract, or clean existing data.
The important distinction is that they’re best suited for deterministic work. You’re giving it information you already have and asking it to perform a predictable operation on it.
For example, you might have a person’s complete work history and want a comma-separated list of every company they’ve worked for. Or you might want to calculate their total years of experience from that same data.
The information is already there. AI Formulas help turn it into the format or value you actually need.
AI Formulas are a good fit when three things are true:
Think calculations, patterns, extractions, formatting, and cleanup.
If you’re asking Clay to research something new on the web or make a more open-ended judgment, you’re moving beyond what AI Formulas are best suited for.
There are plenty of ways to use AI Formulas during the Transform stage.
You could extract all the companies someone has worked for into a comma-separated list, calculate their total years of professional experience, or figure out how long they spent in their previous role.
They can also handle smaller cleanup and formatting jobs, including:
These are the kinds of tasks that might otherwise require custom spreadsheet formulas or scripting. AI Formulas give you a simpler way to apply that logic across the table.
AI Formulas don’t consume Clay credits, which makes them useful for more than convenience.
They also process faster than API-based enrichments, can maintain a consistent output format, and can be reused across different tables and workflows.
That makes the choice fairly simple when you’re transforming data you already have. Use AI Formulas for the structured, repeatable work they can handle, and save your Clay credits for the steps that actually need new information, external data providers, or generative AI.
By the end of Transform, the goal is to have cleaner, more consistent data that’s ready for the final stage of FETE: Export.
At this point, you’ve found the right records, enriched them with the data you need, and cleaned everything up. Export is the last step in FETE, and it’s where that work leaves Clay and gets into the hands of the people and systems that will actually use it.
Depending on the workflow, that could mean sending the finished data to your CRM, Google Sheets, a sequencing tool, a CSV, or another part of your GTM stack.
The first decision is whether you need the data once or need it to keep updating.
For a finished list that only needs to move from Clay to another system one time, a CSV may be all you need. Download the file and upload it wherever it needs to go.
An integration makes more sense when the workflow is ongoing and the destination needs to keep receiving new or updated data.
A simple way to choose is to look at three things:
The answers will usually tell you whether a static export is enough or whether the workflow needs an ongoing connection.
Before exporting anything, clean up what you’re sending.
Your working Clay table may have accumulated a lot of columns along the way. Some were used for enrichment, others for formulas or intermediate steps, and some may have only been needed while you were testing the workflow.
The person receiving the data probably doesn’t need to see all of that.
A cleaner approach is to duplicate your working view and turn the copy into an export view. Hide the intermediate columns and leave only the final fields that need to go downstream.
Your working table can stay as detailed as it needs to be while the export stays clean and easy to use.
For a one-time transfer, CSV is the simplest option.
Once your export view contains only the fields you want, download it as a CSV. That gives you a static copy of the data that you can load into another system.
Clay also keeps an Export History, which gives you a record of previous exports and lets you access them again when needed.
From there, the CSV can be uploaded into the CRM, sequencing platform, spreadsheet, or other system where the team will work with the data.
Just remember that the CSV represents the table at the time you exported it. Changes made in Clay afterward won’t automatically appear in that file.
Google Sheets makes more sense when you want an ongoing connection instead of a one-time file.
CSV = static copy
Google Sheets integration = ongoing connection
With the Google Sheets integration, you can send Clay data into a sheet without repeatedly downloading and uploading CSV files as the workflow changes.
This can also be useful when other people need access to the finished data but don’t necessarily need to work inside Clay themselves. They can use the Google Sheet while Clay continues handling the workflow behind it.
Google Sheets and CSV are only two possible destinations. The finished data can also be sent into the CRM, sequencing tools, or other systems involved in your GTM process.
The destination should come back to what you wanted the workflow to accomplish in the first place. A prospecting workflow may end with contacts ready for outreach. An enrichment workflow may end with updated CRM records. Another workflow may simply need a clean dataset that another team can use.
That completes FETE: Find the right records, Enrich them with useful context, Transform the data into a usable format, and Export it to wherever the next action happens.
A CSV works well when you need a one-time export, but Google Sheets is more useful when the data needs to stay accessible outside of Clay or continue moving as part of the workflow.
Clay can add new rows, look for records that are already in a sheet, and update them when something changes. Which action you choose depends on what you want to happen to the data after it leaves Clay.
There are three main ways Clay can work with records in Google Sheets.
Add Row does exactly what it sounds like. It takes data from a Clay record and adds it to Google Sheets as a new row.
This works well when you know every record being sent should be new and you don’t need to check the sheet first.
Lookup Row searches the sheet for an existing record instead of immediately creating a new one.
This is useful when Google Sheets already contains data and you need to know whether a person or company is there before deciding what should happen next.
A unique identifier makes that lookup much more reliable. For people, that might be an email address. For companies, it could be the company domain.
Lookup, Add, or Update Row combines those actions into one workflow.
Clay first looks for a matching record. If it finds one, it can update that row with the latest information. If there isn’t a match, it can add the record as a new row.
This is especially useful for an ongoing workflow where the same records may come through Clay more than once.
Once you know which action you need, connect your Google account in Clay and choose the spreadsheet you want to work with.
Next, select the specific sheet and map the fields from your Clay table to the corresponding Google Sheets columns. If your Clay table has fields for full name, work email, company, and job title, for example, you can map each one to the column where you want that information to appear.
Once the fields are mapped, run the integration on a small sample first. Check the Google Sheet itself to make sure the data landed in the right columns and looks the way you expected before sending a larger batch.
Be careful about using Add Row by default when you’re working with an existing dataset. If the same person or company comes through the workflow again, you can end up creating another row for a record that’s already there.
Instead, use a field that can reliably identify each record.
Then look for that identifier in Google Sheets before deciding what to do with the record. Existing records can be updated while net-new records can be added.
This keeps one record from turning into multiple versions of itself as the workflow runs over time.
Google Sheets can be a useful destination when the finished Clay data needs to be shared or worked with outside of Clay.
For example, you can use the connection for:
The right setup depends on what happens after the data reaches the sheet. A simple Add Row action may be enough for a one-way list. If Google Sheets is part of an ongoing workflow, looking up and updating existing records gives you much better control over the data over time.
Sending data to your CRM is where you want to be a little more careful than you might be with a CSV or Google Sheet. Once Clay starts creating or updating CRM records, a small mistake can turn into a lot of duplicate or messy data pretty quickly.
Clay supports CRM integrations including HubSpot, Salesforce, Pipedrive, Close, and Copper. The exact setup will vary by CRM, but the basic workflow is the same: check what already exists, decide which records should be added, map the data correctly, and test everything before you scale.
Clay GTM automation can move a lot of records without much manual work. That’s great when the workflow is set up correctly. It’s less great when every row gets pushed into your CRM whether it belongs there or not.
You don’t want the same person created twice because one record used a work email and another used a personal email. You also don’t want every company or contact you research automatically becoming a CRM record.
Be deliberate about what gets added. Your CRM should end up with cleaner, more useful data, not simply more of it.
Before creating a contact, check whether that person is already in your CRM.
Add a CRM lookup to your Clay table and use a reliable identifier to search for a match. For contacts, work email is usually the best place to start because it gives you a fairly direct way to match a Clay record with an existing CRM record.
There will be cases where email alone isn’t enough. Someone may have changed companies, used a different address, or already exist in the CRM under different information. When your workflow requires it, you can add secondary matching logic to account for those cases.
The important part is that the lookup happens before the create action.
Then test it on a small sample. Run around 10 rows that include a mix of people you know are already in the CRM and people you know are new. Check whether Clay is separating the two correctly before you let the workflow go any further.
Once Clay knows which contacts already exist, you can use that result to control who gets created.
Add the appropriate create-record action for your CRM and map the fields from Clay to the corresponding CRM properties. That could include information such as name, email, company, job title, or whatever other fields the workflow needs.
Then add a conditional run so the create action only fires when the lookup did not find an existing record.
Test this with another small batch before running it across the full table. And don’t stop at seeing a successful result in Clay. Open the CRM and look at the records themselves. Make sure they were created in the right place, the fields mapped correctly, and nothing unexpected came through.
The Transform work from earlier in the guide becomes especially important here.
Your CRM has rules about how its properties are formatted, and the data coming from Clay needs to match them. A field that looks fine in a Clay table may not be ready to drop directly into the corresponding CRM property.
Clean and normalize the data first, then make sure each field is formatted for its destination. Company names, phone numbers, locations, dates, and other values should be in the form your CRM expects before you send them over.
You also don’t need to export every column you built in Clay. Some fields exist only to help with research, enrichment, calculations, or decisions inside the workflow. Map the fields that have an actual purpose in the CRM.
This is another reason the FETE order matters. By the time you reach Export, the data should already be cleaned and ready for the system receiving it.
A workflow that works correctly today can still need attention later. Your Clay tables change, CRM properties change, and the way your team uses the data can change too.
For ongoing CRM workflows, keep the same guardrails in place: look up records before creating them, use conditional runs to control when actions happen, and keep records synchronized when the use case calls for it.
It’s also worth checking the destination periodically rather than assuming that a successful Clay run means everything downstream is perfect. Look at the actual CRM records and make sure the data is still landing where and how you expect it to.
The goal is to move the right data into the CRM without creating cleanup work for the people who have to use it.
You’ve now seen each part of Clay on its own. Put them together, and the workflow is much simpler than the number of features inside Clay makes it seem.
Start with a GTM goal, follow FETE from Find through Export, and use the Jigsaw framework to make sure you have the right data before adding more. Here’s what that looks like from start to finish.
Start with what you’re trying to accomplish, not with an enrichment or a Clay feature.
Maybe you want to build an outbound list for a specific ICP, identify companies showing a hiring signal, enrich inbound leads before routing them, or find accounts that meet a very specific set of criteria.
Be clear about the outcome because it determines what you find and which data is actually worth enriching later.
Build the initial dataset around that goal.
For a typical B2B outbound workflow, that usually means finding companies that match your ICP first, then finding the right people at those companies.
Keep the list focused. You’re testing an ICP hypothesis, not trying to build the biggest list possible.
Before moving on, make sure you have your Jigsaw corner pieces. For companies, that means identifiers such as the company domain and LinkedIn company URL. For people, you want identifiers such as full name and LinkedIn profile URL.
Those pieces give the enrichments that follow a better foundation to work from.
Once the audience is in place, add the information that will help you qualify, prioritize, segment, or contact those records.
Company enrichment might include headcount, industry, funding, technology, or hiring activity. People enrichment could add work emails, phone numbers, job history, or other contact information.
Use Clay waterfall enrichment when you want to check multiple providers for a piece of data rather than relying on a single source.
Then use Claygent for the information that isn’t readily available as a standard enrichment field. That could be something as specific as whether a company offers a free trial, mentions a certain technology, or meets a niche condition in your ICP.
You don’t need every available field. If the data won’t change who you target, how you prioritize them, or what you say, it probably doesn’t need to be in the workflow.
Before sending anything downstream, clean up what you’ve collected.
Use Clay’s normalization tools for straightforward cleanup such as company names, whitespace, phone numbers, and locations. Use AI Formulas when you need to extract, calculate, format, or restructure information that’s already in the table.
The finished data should match the format required by whatever comes next, whether that’s your CRM, Google Sheets, a sequencing platform, or another GTM system.
Don’t run a brand-new workflow across thousands of rows immediately.
Start small and inspect what happens. Check that your enrichments are returning the information you expected, Claygent is answering the question you intended, formulas are producing consistent results, and your conditions are firing on the right records.
The same applies to Export. Test the workflow on a small batch and then check the destination itself. If you’re sending records to your CRM, open the CRM and make sure the fields, formatting, and records look right.
Fix problems while they affect a handful of rows, not after they’ve affected hundreds.
Once the workflow is working correctly, send the finished data where it needs to go.
That might mean a one-time CSV export, an ongoing connection to Google Sheets, or sending qualified records directly into your CRM.
For ongoing workflows, use lookups and conditional logic to control what happens downstream. Check whether a record already exists before creating another one, and only trigger actions for rows that actually need them.
A successful run doesn’t mean the workflow is finished. Look at what happened after the data left Clay.
Review:
Use those results to decide what to change. You may need to tighten the audience, change an enrichment, reorder a waterfall, improve a Claygent prompt, remove a field you aren’t using, or test a different signal.
Then run it again.
That cycle of testing, learning, refining, and scaling is what turns a Clay table into a repeatable GTM automation workflow.
The easiest way to approach Clay is through the FETE framework: Find, Enrich, Transform, Export.
Clay is used to build and automate GTM workflows around prospecting, enrichment, research, data cleanup, and activation. Teams can use Clay to find companies and contacts, add missing data, research specific buying signals, prepare records for outreach, and send finished data into a CRM or other GTM system.
Clay enrichment adds structured data such as company size, funding, technology, work emails, and job history to existing records. Claygent is an AI research agent designed to find more specific information that may not exist as a standard database field, such as whether a company offers a free trial or mentions a particular technology on its website.
Clay waterfall enrichment checks multiple data providers in sequence until it finds the requested information. If the first provider doesn't return a result, Clay moves to the next one and stops once the data is found. This can improve data coverage while avoiding unnecessary provider calls.
Start with a specific GTM goal and follow the FETE framework: Find, Enrich, Transform, Export. Find the right companies or people, enrich them with the data you actually need, clean and prepare that data, and then export or sync it to the system where your team will use it. Test the workflow on a small number of records before scaling it.
Test workflows on small samples before scaling, use conditional runs so enrichments only run when necessary, and avoid paying for information you already have. Clay's native normalization tools and AI Formulas can handle many cleaning, formatting, extraction, and calculation tasks without consuming Clay credits.