This is a guide to the GTM (go-to-market) data stack, with every tool turned into a price per row.
The data stack is the set of tools you use to find companies and people, check their emails and reach them by cold email or LinkedIn. A row is one line in your table: one company, one person or one email.
Pricing pages sell plans and credits, but your list is built from rows, so compare rows.
We’ll go through the stack job by job: finding companies, finding people, finding and checking emails, sending, and running it all from Claude Code, Anthropic’s AI agent that runs in your terminal.
One example runs the whole way through: a list of US commercial cleaning companies, with an owner’s email for as many as possible.
In the example, 10,000 companies that fit become 3,770 owner emails you can send to. The data costs $216. The first month, with sending, costs $511.
This guide is for founders and GTM engineers who buy their own data. If you already have a list of companies, skip to step 2.
How to price a tool per row
Price per row = what you pay / rows you use. Count the rows you use, not the rows your plan includes. If a plan gives you 5,000 credits a month and you use 2,000 before they reset, your real price per row is 2.5 times the one on the page.
A credit is the unit a tool bills in, and each action costs a set number of credits. Before you compare tools, convert every credit price into dollars per row.
Then look at how the tool charges. That decides where the tool goes in your pipeline, the order in which your list moves through the tools:
- Per row or per search, even when the result is useless. Maps searches, scrapers (tools that copy data from websites) and email verifiers charge for every row or search. Send them as few rows as possible.
- Per hit. Most email finders charge only when they find a valid email. A miss costs nothing, so you can run several of them one after the other.
- Flat rate. One payment a month for the whole database, either unlimited under a fair-use policy (the tool may limit you if you pull far more than normal) or up to a set number of records a month. The more rows you pull, the lower your price per row.
Also check what happens to credits you don’t use. Prospeo’s credits reset every month. LeadMagic’s carry over until you hold twice your monthly amount, and MillionVerifier’s never expire. Credits that reset make a big plan expensive for a one-off list.
1. Find the companies
Where you collect companies depends on your market. Here are the common sources and what one company costs on each.
Companies on LinkedIn (software, services, agencies):
- Prospeo: searching is free, and exporting a company costs 1 credit. That’s $0.0245 per company on Starter ($49 for 2,000 credits a month), $0.0198 on Growth ($99 for 5,000) and $0.0166 on Pro ($249 for 15,000).
- BlitzAPI: flat rate with a monthly limit. Start costs $99 a month for up to 200,000 records. A record is one company, person, email, phone number or job post it returns, and a search that finds nothing counts nothing. Scale costs $499 for 5 million.
- GetLeads: flat rate. Unlimited costs $497 a month for its whole database of people and companies.
- Apollo is another option. Each plan comes with a pool of credits.
For a flat plan, divide the monthly price by the rows you pull. $99 divided by 10,000 companies is $0.0099 per company. If you use all 200,000 records, it’s $0.000495.
Local businesses (trades, cleaning, repair, anything with a storefront). Most of them are on Google Maps, and many have no LinkedIn page. You can get Maps data two ways:
- Maps Data on RapidAPI. RapidAPI is a marketplace for APIs, which are tools your script calls to get data back directly. Maps Data has a free plan with 1,000 searches a month. Paid plans cost $5 for 30,000 searches or $25 for 300,000 searches a month, and each search returns up to 200 businesses.
- Google Maps Scraper on Apify. Apify is a marketplace for scrapers. This one costs $4.00 per 1,000 businesses on the free plan, $3.00 per 1,000 on Starter ($19 a month) and $1.50 per 1,000 on Business ($999 a month).
The two charge differently. On the $25 plan, one Maps Data search costs $0.0000833, whether it returns 200 businesses or 3. Apify charges $0.003 for every business it returns on Starter. So per business, Maps Data is cheaper as long as you use the searches you pay for.
Maps only shows businesses near the place you search, so you search ZIP code by ZIP code. My TAM playbook for local businesses walks through that search. A TAM (total addressable market) is every company that could buy from you.
Online shops. Store Leads is a database of online shops. Its Pro plan costs $250 a month and includes exports. The $75 Premium plan doesn’t.
Niches without a clear label. A lookalike search finds companies similar to examples you give it. Prospeo includes AI lookalikes in its paid plans. Exa, an AI search engine, charges per request, from $4 per 1,000 searches. Ocean.io is a third option.
Free sources. Your own CRM (the software where you track customers and deals) costs nothing, and neither do public registers for licensed or regulated companies. Job posts are often free too. Greenhouse and Lever, two common hiring systems, publish each company’s open jobs through a public API that needs no key.
2. Find the people
Next you need a named person at each company, for example the owner or the head of sales.
If you build lists every month, start with the flat-rate tools. GetLeads and BlitzAPI both charge one payment a month: GetLeads for unlimited access under a fair-use policy, BlitzAPI for up to 200,000 records on Start. Use Prospeo last, because you pay per contact there: 1 credit for each person it finds with an email, and nothing when it finds none.
We pay BlitzAPI a flat monthly fee ourselves, because we build lists every week.
If you build one list, compare one month of a flat plan with paying per contact. For 2,000 contacts once, Prospeo’s Starter plan costs $49, and one month of BlitzAPI Start costs $99.
Local businesses are different. Many owners of small shops have no LinkedIn profile, so tools built on LinkedIn data find fewer of them. A web search for “who owns [company] in [city]”, read by an AI model, often finds more. My TAM playbook compares these sources: LinkedIn databases, web search, state business registers and the company’s own website.
3. Find the emails, then check them
Guess first. If you have the person’s name, try the common formats, like first.last@company.com. Then check each guess with a verifier, a tool that checks whether a mailbox exists without sending an email. That costs only the check.
Then pay an email finder for the rest. These three common tools all charge only when they find a valid email:
| Tool | Entry plan | Per found email | Larger plan | Per found email |
|---|---|---|---|---|
| Prospeo | Starter, $49 for 2,000 credits | $0.0245 | Pro, $249 for 15,000 | $0.0166 |
| LeadMagic | Essential, $99 for 5,000 credits | $0.0198 | Ultimate, $849 for 100,000 | $0.0085 |
| Icypeas | Basic, $19 for 1,000 credits | $0.019 | Hypergrowth, $499 for 100,000 | $0.00499 |
Icypeas adds VAT where it applies, and its credits never expire. BlitzAPI includes verified work emails in every plan, inside the monthly record limit.
Run the finders as a waterfall. A waterfall means you run finder one on all rows, send only its misses to finder two, and stop at the first valid email. You pay per hit, so you never pay twice for the same row. Test every finder on the same 200 rows from your market before you pick the order. In my experience, the hit rate (the share of rows where a tool finds an email) changes a lot between markets.
Verify every email before you send. Send only to the addresses the verifier marks valid.
MillionVerifier sells packs that never expire: 10,000 checks for $39 ($0.0039 each), 100,000 for $149 and 1 million for $449 ($0.000449 each).
Check catch-all addresses with a tool built for them. Some mail servers accept mail for any address, so a normal check can’t confirm the address. These are catch-all addresses. BounceBan charges $34 for 10,000 checks ($0.0034 each). LeadMagic’s email check costs 0.25 credits, $0.00495 on Essential. Send to a catch-all address only after one of them confirms it.
Drop every address that comes back “unknown”, because the verifier couldn’t confirm it.
Buy phone numbers only for the companies that fit you best. A mobile number costs 5 credits on LeadMagic ($0.099 on Essential) and 10 credits on Prospeo ($0.198 on Growth). BlitzAPI includes US phone numbers on every plan, inside the record limit, and GetLeads includes mobile numbers in its plan.
4. Choose a sending tool
The sending tool is a small part of the bill. Mailboxes and domains cost more.
- Instantly: Growth costs $47 a month for unlimited email accounts, 1,000 contacts and 5,000 emails. Hypergrowth costs $97 for 25,000 contacts and 125,000 emails, so $0.00388 per contact if you use all 25,000.
- Smartlead: Base costs $39 a month for 2,000 contacts and 6,000 emails. Pro costs $94 for 30,000 contacts and 90,000 emails, so $0.00313 per contact at full use. Smartlead also sells Google mailboxes for $4.50 per mailbox a month, plus $13 per domain a year.
- HeyReach sends on LinkedIn. Growth costs $79 per LinkedIn account a month. Say one account reaches 400 new people a month. Then each person costs $79 / 400 = $0.1975.
That puts a person reached on LinkedIn on the most expensive level, next to mobile numbers. I’d keep LinkedIn for the companies that fit best.
5. Run it from Claude Code
Claude Code writes the scripts, calls each tool’s API and keeps a cost log. It comes with Claude Pro at $20 a month, and Max starts at $100 a month if you need more usage. You install it with one command and start it by typing claude in your project folder. There’s a separate setup guide for that.
Put your API keys in one file. An API key is the password a script uses to call a tool. Create a file called .env in the project folder with one line per tool, like PROSPEO_API_KEY=.... The scripts read the keys from there, and the keys never go into the chat.
Use it for the free steps. Removing duplicates, filtering by category and checking websites cost nothing when a Python script that Claude Code writes does them. We use Claude Code together with Gemma, a free AI model from Google that runs on your own computer, to read company websites and decide which companies fit. Before the full run, test the question on a few dozen companies you labelled by hand, and fix it until the model agrees with you.
Some websites block scripts. We open those with Playwright, a free tool that lets a script open pages in a real browser.
Connect the tools. For bulk jobs, Claude Code writes a script that calls the tool’s API with your key, and the results go into a file. For quick questions, you can add an MCP server, a connector that lets Claude Code call a tool directly in the conversation. Prospeo, LeadMagic, Icypeas, GetLeads and HeyReach all offer one.
Keep every price in one file. Here’s the price file I’d start with. Replace the plans with your own:
# prices.yaml: price per billed unit, USD
maps_data_search: {billing: per_search, price: 0.0000833, plan: "$25 for 300,000"}
apify_maps_place: {billing: per_row, price: 0.003, plan: Starter}
prospeo_company_export: {billing: per_row, price: 0.0198, plan: Growth}
prospeo_email: {billing: per_hit, price: 0.0198, plan: Growth}
icypeas_email: {billing: per_hit, price: 0.019, plan: Basic}
millionverifier: {billing: per_row, price: 0.0039, plan: 10K pack}
bounceban: {billing: per_row, price: 0.0034, plan: 10K pack}
blitzapi_start: {billing: flat, price: 99, per: month, records: 200000}
getleads_unlimited: {billing: flat, price: 497, per: month}
And the prompt that adds a cost log to your pipeline:
Add a cost ledger to this pipeline. Read prices from prices.yaml, never hardcode a price.
For every step log: step, tool, plan, billing type, rows in, rows billed, rows out, cost.
For per_search tools, count searches, not rows.
After each run print the total cost, the cost per surviving row, and every step where
rows billed is more than twice rows out. Those steps need a free filter in front of them.
6. Pay for the cheapest level first
Sort every tool into four price levels, from free to most expensive. I call this the cost ladder. Start at the cheapest level, and send only what’s left to the next one.
Between the levels sit free filters: remove duplicates, check the category, read the website, remove customers, open deals and everyone who asked not to be contacted. Every row a free filter removes is a row you never pay for later.
The four levels:
- $0: your CRM, public registers, public job boards, and Claude Code with Gemma reading websites.
- Under one cent per row: Maps searches, Maps businesses on Apify ($0.0015 to $0.004), email checks ($0.000449 to $0.00495).
- One to three cents: each email a finder finds, or each company record, on the entry plans of per-credit tools ($0.019 to $0.0245).
- Four cents and up: mobile numbers ($0.0425 to $0.245) and each person you reach on LinkedIn.
Flat plans are not on the ladder. Their price per row depends only on how many rows you pull.
Three rules go with the ladder:
- Check fit before you pay for contacts. Category, website text and size are free to check. Do it before any row reaches a paid finder.
- Remove duplicates by what the tool charges for. Two Maps listings with different names but the same website are one company. Remove duplicates by website, or you pay twice for that company.
- Count searches for tools that charge per search. A search that returns 3 businesses costs as much as one that returns 200.
7. Worked example: 10,000 companies end to end
Say you sell a scheduling tool to US commercial cleaning companies. You want every company in the market and an owner’s email for as many as possible. The prices below are real, and the counts and hit rates are an example. For scale, the Census counts 67,799 US janitorial service locations with paid employees in 2023, home cleaning included.
| Step | Tool, plan | Rows in → out | Billed | Cost |
|---|---|---|---|---|
| 1. Search Maps for 3 terms in every ZIP code | Maps Data on RapidAPI, $25 plan | 115,824 searches → 48,600 businesses | 115,824 searches | $9.65 |
| 2. Remove duplicates and wrong categories | Python script | 48,600 → 29,300 | 0 | $0 |
| 3. Read every website, keep the ones that fit | Claude Code with Gemma | 18,100 with a website → 10,000 | 0 | $0 |
| 4. Find owners with Prospeo’s people search | Prospeo, Growth | 10,000 → 2,900 owners with email, 1,600 names without | 2,900 | $57.42 |
| 5. Guess 3 email formats for the 1,600 names | MillionVerifier, 10K pack | 4,800 guesses → 700 valid | 4,800 | $18.72 |
| 6. Email finder on the 900 names left | Icypeas, Basic | 900 → 400 found | 400 | $7.60 |
| 7. Verify the found emails | MillionVerifier, same pack | 3,300 → 2,880 valid, 320 catch-all, 100 invalid or unknown | 3,300 | $12.87 |
| 8. Check the catch-alls | BounceBan, 10K pack | 320 → 190 valid | 320 | $1.09 |
Step 1 is 3 search terms times the 38,608 US ZIP codes with residents. In step 3, the 11,200 businesses without a website don’t get a website check. You check them by name and Maps category and keep them as a calling list.
In step 4, Prospeo charges only for owners it finds with an email, so the 1,600 names without an email cost nothing. For the 5,500 companies where no owner turns up, the “who owns” web search from step 2 is the next try.
Usage comes to $107.35, with Maps counted per search. What you buy is $216: the $25 Maps plan, Prospeo Growth at $99, the MillionVerifier pack at $39, Icypeas Basic at $19 and the BounceBan pack at $34. At this size, the plan prices matter more than the prices per row. That’s why step 6 uses Icypeas: 900 lookups don’t need a $99 plan.
Steps 2 and 3 cost nothing. They kept 38,600 businesses away from the paid tools, and 11,200 of those stay on the calling list.
The 2,100 Prospeo credits you didn’t use are lost at the end of the month. The MillionVerifier and BounceBan credits you didn’t use stay.
Then sending. 3,770 contacts don’t fit into Smartlead Base, so you need Pro at $94 a month.
Cold email is spread over many mailboxes on extra domains, so each mailbox sends only a few emails a day. Say you run 20 Google mailboxes on 7 domains: that’s $90 a month, plus $91 for the domains, paid up front for the year. Add Claude Pro at $20.
First month in total: $511. That’s under 14 cents per contact you can send to. The data alone is under 6 cents per contact.
Phone numbers raise the bill the most. At 5 LeadMagic credits each, mobiles for the 300 best-fit companies take 1,500 credits, covered by Essential at $99. All 3,770 contacts need Growth at $249.
Compare a flat plan if you build lists every month. For companies on LinkedIn, flat-rate tools find most of the people. BlitzAPI Start costs $99 for up to 200,000 records a month, emails and US phone numbers included. GetLeads costs $497 with no record limit, so $0.0497 per person at 10,000 people or $0.00497 at 100,000. For local businesses like the ones in this example, they find fewer owners, so test them on 200 rows first.
Avoid two pricing mistakes
- Reading the “from” price as your price. “$1.50 per 1,000” on the Apify Maps scraper is the Business plan at $999 a month. On Starter it’s $3.00.
- Using old prices. Prices change. Check the pricing page before you buy, and keep prices in a file instead of in your code.
Where to start today
You can price your own stack per row in three steps:
- Write one line per step of your pipeline: tool, plan, how it charges, rows billed and rows kept.
- Divide what you paid by the rows you kept. That’s your price per row.
- Put a free filter in front of the step where rows billed is far above rows kept. That’s where your money goes.


