How to Refer Data-License Deals, and What Changes If You Become the Broker

To refer a data-license deal, get the company's consent and connect it with a buyer for its own operating record. Becoming the broker adds qualification and coordination, without taking custody of the files. I would count replies and booked calls separately from signed licenses or payouts, because our campaign file proves only the former.

5,078 Google-hosted contacts produced 27 engaged people and 6 booked calls. The same copy, on a separate other-host list, bounced 39% of the sends.

We ran those campaigns for a data buyer. On the whole file, through 2026-10-09, there are 280 reply rows, 45 engaged prospects, and 10 booked meetings. Meeting outcome is blank on 9 of the 10, and there is no offer, intake, or payout on file. Every time I say something worked, I mean a person replied or booked a call. I do not mean a company signed a license.

Treat the $50k video as his attended-call claim

Eric Nowoslawski posted "I speedran cold email from 0 to $50k in 30 days" on YouTube on 2026-09-10.

https://www.youtube.com/watch?v=-LjvwKOIH9Y

The counts below are from the auto-generated English transcript. I have not checked them against the audio.

He says his agency, Growth Engine X, sent over 100,000 emails in less than 30 days, about 106,000, so they could collect $50,000 as an affiliate. He says one positive response for every 286 emails, 373 positive replies, about a 62% booking rate, 231 meetings, and that they were paid per attended call. He says that amounted to $50,000. He also says the first half has been paid and the second invoice is already out. He says they never do performance deals, and that he took this one so they could be part of the upside.

The description on the same page says they found someone willing to pay a success fee as a percentage of the data contract, and it points at Polyshares for the offer details. I am not folding those two sentences into one pay structure. The spoken $50,000 is tied to attended calls. The description talks about a percentage of the contract. I have not seen a contract, an invoice, or the sending account.

Those counts are consistent with each other. 373 times about 62% is about 231, and about 106,000 divided by 286 is about 371, next to the 373 he states. That scoreboard is not ours. Our file has 10 booked meetings and no payout. His word is "positive." On our best Google campaign, 16 of 5,078 were marked interested, about 1 in 317, beside his 1 in 286. Same neighborhood, different definition, and still a reply rather than a payment. Do not put his $50,000 in an email.

Keep Google mail and drop the rest

Google-hosted mail replied. Microsoft-hosted mail and the other-host file did not.

The best single campaign was a mixed list of small firms on Google-hosted mail. We loaded the copy in three frames: "we pay $150k to $1M", "exit money", and "not a sale, you keep the firm", plus a step-2 line about tenure. Of 5,078 people contacted, 27 engaged and 6 booked. Per 1,000 contacted, that is 5.3 engaged and 1.2 booked. 16 were marked interested, which is 3.2 per 1,000. Those 16 split as law 7, architecture 3, B2B data or tech 4, accounting 1, and other 1.

The same copy on a list of other-host mailboxes was a list failure. We loaded 5,309 and sent 249. 98 of those sends bounced, 39%. The export shows 1 engaged person and 0 booked calls.

Microsoft-hosted recipients in campaigns 69 through 87 produced 0 human replies out of 895 contacted. A separate diagnostic in the same notes puts Google-hosted recipients at a 1.84% reply rate. That 1.84% is a different cut from the 5.3 per 1,000 on the 5,078-person campaign, so I would not average them. The rule for the next list is concrete: keep Google-hosted mail, strip security-gateway mailboxes, and do not read a Microsoft-hosted silence as a verdict on the words.

*Results by mail host, campaign file through 2026-10-09. No offer, intake, or payout is recorded for these campaigns.*

BUILD THE ROW
Open Places lists, united with Google Maps pulls
reviews 20+ | domain age 3+ years | size gate
        |
        v
KEEP OR DROP THE MAILBOX
Google host --> keep
Microsoft or other host --> drop
security gateway --> strip
        |
        v
WRITE EMAIL 1
you keep the firm | three reasons from the public site | one small ask
        |
        v
CLASSIFY THE REPLY
what data | legal or confidential | who are you | how much | a call
a reply or a booked call is the result on file
        |
        v
STOP BEFORE A PAYOUT CLAIM
no offer, intake, or payout is recorded for these campaigns

License the company's own record

A data license, in the campaigns we ran, is a buyer paying a company for its own operating history. That means job notes, estimates, tickets, project files, and the internal trail of how the work got done. The company keeps operating. Client files are a different pile. Polyshares says material governed by a client contract belongs to the client, so those files are out. I do not have a public micro1 page that repeats that line.

Polyshares publishes that it licenses the operating record a business produces by running, including Slack, email, tickets, CRM, and files, and that typical deals run $100K to $2M and above. That range is the buyer's own line. On its track-record page it also publishes four licenses, with the company names kept confidential: a roofing company at $450,000, a general contractor at $300,000, a solar company at $275,000, and a global marketing agency at $110,000. Each is described there as a perpetual exclusive license, anonymized before delivery. All four amounts are buyer-stated. The homepage, not the track-record page, says the partners previously built a roofing company and a solar company. The track-record page does not say the $450,000 or $275,000 licenses are those firms. Do not treat those two licenses as the partners' own firms.

Polyshares also publishes that the length of the record matters more than headcount, that nothing leaves the company's systems until an offer is accepted, that intake to an offer is 1 to 2 weeks, and that payment is net 30 to 60 after the data is approved, anonymized, and delivered. Those are the buyer's claims about its own process. I do not have a matching payout in our campaign file to set beside them.

micro1 publishes a commitment to spend $1B over the next 12 months acquiring and licensing enterprise operational data, and a partnerships page aimed at companies with 30 or more employees, established documentation, and operational data such as written procedures, knowledge bases, CRM data, project histories, and internal processes. The public referral page prioritizes the United States. micro1 has also said companies keep ownership of the underlying data. The partnerships page lists price tiers of $100k+, $500k+, and $1M+. Its referral page says a referrer can earn up to $50K per referral, paid after the referred company clears the threshold that page describes. That "up to $50K" is buyer-stated. It is not a fee I have received, and it is not a number to promise a founder.

Spirit Airlines is the one operating-record case where the amounts come from a court record, and it is not a small-company sale. The bids are not all micro1's. Mercor was the alternate bid at $7.5M. Google won the August 2026 auction at $10M. micro1 later noticed a competing bid at $12.5M. As of 2026-10-10 the sale was still not closed, and a hearing was reported for 2026-10-14. Do not show a roofing firm the $12.5M as its comp.

One record goes to one buyer. The four published Polyshares licenses are described as perpetual and exclusive. A separate Polyshares answer also describes a bounded non-exclusive license, so do not tell a founder the license is exclusive unless that offer says so. The micro1 terms I am using have the seller warrant that it has not already granted someone else the same kind of rights. Do not submit the same company to both.

I am an independent referrer. I am not micro1's agent, recruiter, representative, partner, or spokesperson. Both links in this piece are referral links.

Polyshares, for a company's own operating record: https://polyshares.com/signup?invite=a4de836f166726d27c21f826d5da9b6e

micro1, for a company with 30 or more staff and its own internal record: https://refer.micro1.ai/register/?affiliateCode=f6ad03af-a739-43c1-bab1-5ac52039e45a&utm_source=referral&utm_medium=share&utm_campaign=job_referral

The referral terms I am using are dated 2026-10-08. I could not re-check them on a public page. They require consent from the company before you refer it, require the company to be new to them, and require the outreach to be truthful. The terms do not say cold email is allowed, and they do not say it is banned. Until that is explicit, I would not start a cold sequence that names micro1. I would send the link when the company asks, or after a person has said yes to being referred. The terms also say you do not share the referral payout with the company you referred.

Rank who can say yes, and park who only replies

Law produced the most interest on the best campaign and is the wrong list. Client-held files are outside what Polyshares says it will license. The micro1 terms I am using exclude payment-card and financial-account data, health records, biometric data, children's data, export-controlled data, scraped data, and AI-generated content, except where an order form says otherwise. Accounting has the same problem on client work. On the small-firm accounting sends, interested was 0 and booked was 0. A later accounting reload of 2,673 leads produced 1 engaged person and 0 booked calls, and contacted volume for that reload is not on file, so I will not invent a rate.

The ranking below uses published buyer amounts for the money, and our replies only as a sign of interest. A sellability score fitted on 80 labeled firms scored 0.455 against positive replies. A coin flip would be 0.5, so the score does not predict who replies. I use it only to order who might be sellable.

Seven next-campaign ideas were scored from 1 to 5 on six factors, 30 points max, on two separate passes. Trades without a tool requirement scored 23 and 21. Trades with a field-service record scored 22 and 20. Founder-era firms scored 21 and 19. MSPs scored 20 and 19. Scarce-record AEC specialists scored 19 and 20, and most of that list is already loaded. Accounting scored 14 and 13. A warm follow-up to people who had already replied scored 19 and 22, and that pool is small. Those scores order the work. They are not reply rates.

Start with trades that already run a job system. Polyshares' published roofing and solar licenses are the closest public comps, and they are buyer-stated. The company's own job record is the asset. A broad trades starting estimate sat at 0.25, and trades with a field-service tool sat at 0.31, each on 15 labeled firms. Those are starting estimates, not close rates. A trades-owner campaign in October booked 1 meeting, and the send count is unknown, so that booking has no rate. The list already built for the next trades send is 5,128 Google-hosted firms, with about 1,100 verified owner contacts after qualification. Both counts are estimates from the build notes.

Put builders and AEC firms with long project files second. The general-contractor license of $300,000 is buyer-stated. Architecture firms engaged more often than builders in the live AEC file, 7 of 1,978 firms against 2 of 3,766. That is about 0.35% against 0.05%. Treat it as a lead. The holdout in the same test contained 1 positive in total, under the floor of 5 the method requires, so it did not confirm. Campaigns launched 2026-09-30 split the same direction on counts, with no contacted denominator: architecture 7 engaged and 2 booked, engineering 2 engaged and 0 booked, builders 2 engaged and 0 booked. A lot of that AEC list is already loaded. New volume waits.

Park MSPs third until a license amount exists. No published MSP license amount. The playbook records IT as 2 of 5 bookings, and it also parks specialist research and evaluation firms at 2 of 5 bookings because ownership of their reports is often unclear. I cannot tell from the notes whether those are the same two meetings, so I am not treating "IT replies" as a proven segment. The reachable Google-hosted MSP slice is about 600 firms, an estimate, and a lot of this vertical is Microsoft-hosted, which we already saw go quiet.

Refer B2B software only after the company says yes. Fourth, and only as a micro1 referral, not as a cold sequence, for the consent reason above. micro1 is the buyer that publishes a referral amount. The public software comps I found were closed companies, relayed by others, so I would not talk to a running software firm as if those comps were its price.

Use closing companies fifth, and confirm the closure before you write. Broker-stated figures for closed startups sit around $10,000 to $100,000. The seller has a reason to look, and the amounts are small. Confirm the closure from a filing, a WARN notice, or the company's own shutdown post. Do not guess that a company is closing.

Founder-era firms, 30 or more years from a year stated on their own site, are a sort key across the first three segments. The starting estimate there was 0.33 on 15 firms. It is not its own campaign.

Firm size is a weak lead, not a rule. On the AEC lists, firms with 1 to 4 staff engaged at 0.41%, and firms with 50 or more staff engaged 0 of 476. The method marked that weak. I would still email a 60-person firm whose record is long. I would not build the first wave there.

Build the row before you buy a contact

Start from Open Places business lists united with Google Maps pulls. On the accounting and early AEC builds, the gates were reviews of 20 or more, a domain age of 3 or more years, a size gate, and security-gateway mailboxes stripped. A later note prefers 10 or more years because the buyers talk about record depth. That preference is an inference. The sends on file used 3 or more years.

Then look up the mail host. Keep Google. Drop Microsoft and other hosts. Strip security gateways.

For trades, the build notes use roofing, HVAC, plumbing, electrical, and solar, in the US, with an employee range of 11 to 500, and they exclude restoration, landscaping, paving, lead-gen sites, franchise units, and private-equity platforms. Confirm a job system with a first-party signal before you name it. Homepage regex found a vendor on 3 of 150 trades sites and on 30 of 414 AEC and builder sites. DNS, a portal link, or the company's own job post is the check. A technology database's start date missed 78% of firms that were already 10 years old, in the same notes, so do not let that date veto a row.

For builders, use Maps in overlooked metros plus users of project systems, and read the homepage for specialist work such as forensic, facade, envelope, historic preservation, geotechnical, or condition assessment. Pull net-new names only. The September architecture, engineering, and builder campaigns already hold most of this market.

For MSPs, look for the MSP category, then boost rows that show a help desk or documentation tool by a first-party signal. Keep US delivery. A reseller with no ticket history of its own fails the ownership test even if the headcount looks right.

For the micro1 software lane, once consent is settled, filter to software or IT services, headcount 31 to 500, headquarters in the US, UK, or Canada, founded in 2020 or earlier, and an English site. Exclude health-record products, payment and banking data, defense work, scraped-data vendors, and AI-content shops. Do not reuse a company already sequenced for the other buyer.

Titles to find: owner, founder, president, or CEO under about 50 staff. At 50 to 500 staff, the CEO or COO is the person who can say yes, and an operations or IT lead is the person who can count what is in the system. Counsel reviews.

Projection: how many US firms might qualify. The Census Bureau's 2022 Statistics of US Businesses gives a size-and-industry screen of about 29,000 to 44,000 employer-firm entries for roofing, electrical, plumbing, and HVAC in the 10-to-499 and 15-to-499 staff bands, which are coarse proxies for 11 to 500. That is my planning range for the Polyshares trades route before the remaining gates, with solar missing from the industry cut. Public Google-mail and review-count shares measure different populations, so I cannot use them to narrow it. For the broad micro1 referral, the Census counts about 438,000 employer firms with US establishments and 30 or more staff. It does not establish which of them have US headquarters, and no public share supports a further cut across all industries. Within its software and IT lane, the screen is about 9,000 to 10,500 in the 30-to-499 band, a coarse proxy for 31 to 500. The available firm-age share uses wider sectors and size bands, so I cannot use it to narrow that range either. These are projections from size-and-industry screens, not counts of firms passing every gate. Mail host, reviews, domain age, record ownership, consent, and excluded data categories still need a company-by-company check. The sources and working are in the projection notes. No deal count or dollar figure follows from these firm counts.

*Size-and-industry screens from the US Census Bureau's 2022 Statistics of US Businesses. Not counts of firms that pass every gate.*

Keep a generic inbox if that is the only door. 0 of 45 engaged prospects wrote back from a role mailbox. At firm level, 2 of 1,336 firms with a role mailbox engaged, against 9 of 5,311 firms without one. That is 0.15% against 0.17%. Owners answered from their own address after a generic inbox was on the list. Dropping generic-only firms would have dropped those companies. Find the owner. Do not expect info@ to be the person who talks.

Tenure comes from a year the company states on its own site, or you leave it out. Registry and archive dates understated the founding year on 26 of 37 firms checked. An uploaded tenure column was joined to the wrong rows, and it had no relationship to the year on the site. Site-stated years showed up on 28% of AEC sites, 15% of accounting sites, 9% of MSP sites, and 8% of trades sites. When the year is missing, send a version of the email that does not mention years. When you have a dated job post, write "since at least" that year. Do not invent an exact count.

Use the score to order the first wave

The 100-point rubric is a way to sort a list. It is not validated against a sale, and the same project already found that nothing on a homepage predicted replies. A low score is not a reason to delete the row. A high score is not a payout.

Give up to 40 points for the system class. A help desk, field-service platform, project platform, or issue tracker scores 40. A CRM scores 25. A wiki or internal chat scores 15. Accounting or ERP only scores 10. Nothing detected scores 0. Give up to 30 for years in use, using the earliest dated evidence as a floor: 30 at 10 or more years, 24 at 7 to 9, 18 at 5 to 6, 10 at 3 to 4, and 5 if the tool is present but undated. Give up to 15 for evidence strength: 15 when two sources agree and one is first-party, 8 for one first-party source, 3 for a database tag alone. Give 5 points per extra system class, up to 15.

Before you score, apply the gates. Google-hosted mail for the operating-record buyer. 30 or more employees and a US, UK, or Canada headquarters for the micro1 referral. Drop anything in the excluded categories above, for either buyer.

The worked example in the sourcing notes is a roofing company with a field-service tool seen in DNS and on a portal link, a 2017 job post that names it, and a CRM also detected. That scores 89. It is an illustration of the arithmetic. It is not a company we emailed, and it is not a result.

Send 75 and above in the first wave if you need a cut. Hold rows under 50 until the first wave has a reply rate. Bands of 50 to 74 wait in the middle. If the first wave's replies ignore the score, throw the score out. The homepage result says that is a real possibility.

Safe lines in the email are a founding year quoted from the company's site, a vendor name backed by a first-party signal, and a work type quoted from the company's own services page. Employee counts, the rubric score, and a database tag you could not confirm stay out of the copy.

Send the reasons email, and do not borrow the $898K line

A community template set at skool.com/databroker is the version of this offer that gets copied. I am quoting it as their template. It is not our campaign and not our result.

Variant A:

{{firstName}},

We just licensed a {{prospects_vertical}} data for 898K.

That revenue went straight to their bottom line, no borrowing or selling equity.

If we could do the same for {{company_name}}, would that be worth a conversation?

{{accountSignature}}
YourAdvisoryName

P.S - If you're not interested, a simple "nope" works for us.

Variant B adds "If we could license {{company_name}} internal-data for 6-7 figures" and the line "Our buyers can close in under a week." Variant C is that 6-7 figures question with the $898K story removed. I am not adopting the $898K figure, the under-a-week close, or any buyer count attached to that community. We have no recorded payout to put in that sentence. Their usable note is mechanical: do not bury the offer, and if you cite a figure, cite one you can point at for that row. We cannot point at a payout. We can point at a year on the site and a system we saw.

What we tested, on the Google campaign, was three frames in email 1 and a tenure line in email 2. "Not a sale, you keep the firm" drew interest on 7 of 1,739 sends, 0.40%. The base "we pay" frame drew interest on 3 of 1,660, 0.18%. On these counts the gap is about a 1 in 5 chance of showing up even if the two frames were the same, so I would not call the gap real. Reply counts on the three arms were 67 on the base and 64 and 49 on the other two. The notes do not say which of those other arms had 64. I still open from "you keep the firm" because 0.40% was the higher interest rate, and because the sentence matches a license rather than an acquisition. I do not call it a winner.

The step-2 tenure bump produced 0 interested replies out of 1,071. All 16 interested replies on that Google campaign came from step-1 variants. The test does not separate "tenure was the wrong sentence" from "any second email failed," because no other step-2 copy was in the test. I would not send a follow-up whose only new fact is "after N years."

The price band in the "we pay $150k to $1M" opener was a tested frame, and it still needs the buyer's sign-off before a new send. It is not the number on the public site. For a new email I would quote the buyer's published range, say the buyer stated it, and say the reader's number will differ. I would not promise $150k, $1M, or $898K.

Send the reasons email first, with one small ask. Subject, two to five lowercase words, for example "not a sale" or "old job files". About 50 to 90 words. One ask a person can answer in a few words. No calendar link in this email.

{FIRST_NAME}, this is not a sale of {COMPANY}. You keep the firm.

AI labs license a company's own operating record. I refer owners to one buyer. Three reasons {COMPANY} might fit.

1. {TENURE_YEARS} years on file, from your site.
2. {DATA_LINE}
3. {PROJECT_LINE}

The buyer publishes four licenses from $110,000 to $450,000, and says a typical deal is $100K to $2M. Those figures are the buyer's, and yours can differ. The buyer says the company pays no fee.

Want the two-minute check?

Use {DATA_LINE} only when a first-party signal supports it, such as "ServiceTitan job notes, estimates, and call recordings" if DNS or their own portal shows ServiceTitan. Otherwise use a generic line you could defend from the trade itself, such as "job notes, estimates, and change orders." {PROJECT_LINE} is a work type from their services page, such as "Commercial re-roofs and storm claims take real judgment." If you do not have a site-stated year, delete reason 1 and write "Two reasons."

For an MSP, swap the record and say client systems stay out. Do not attach the roofing $450,000 to an MSP. No MSP license is published. The money line I would send is the buyer's general range, $100K to $2M, labeled as the buyer's typical deal, not as their offer.

For a builder, the record is RFIs, change orders, schedules, and field notes. Client-owned drawings stay out. The buyer-stated comp, if you use one, is the general contractor at $300,000, with the same "yours can differ" sentence.

Email 2, two days later, only if email 1 got nothing. The two-day gap is the draft contract for unsent follow-ups, not a measured lift. The body has to add the scope rule, because that is the question engaged people asked. Do not repeat the tenure line.

{FIRST_NAME}, the record in question is {DATA_LINE}. Files you hold for clients stay out.

The buyer says nothing leaves until you accept an offer. I can send the one-page version of what counts.

Want that?

On the early AEC test, the "we pay" arm had 0 human replies on 234 sends, "exit money" had 2 human replies on 239, and "not a sale" had 2 human replies on 259. Those three arms sum to 732 sends. The same note also says 914 sends, which does not match, so I am not using 914 as the base. Interested was 0 and booked was 0 as of 2026-09-29. I am not treating a 2-reply arm as a winner. The sample was too small to rescue a bad host mix.

Answer what the data is before you ask for the call

Of the 45 engaged prospects, 17 asked what the data was, or asked for more detail, in the first reply. 10 asked for a call. 7 raised legal or confidentiality. 3 asked who we were. 3 asked how much. Those counts overlap. They are keyword counts, not a coded study.

36 of the 45 never booked. 41 of the 45 engaged threads end on our message. We have not run a clean test of a better rescue email. The reply I would send next answers the question they already asked, in this order.

If they ask what the data is, name the system and the artifact. Job notes, estimates, change orders, tickets, runbooks, RFIs, schedules. Say that client-owned material stays out, and that the buyer describes delivery as anonymized. Offer the short list of what counts. Do not attach a new dollar figure.

If they ask whether it is legal or confidential, tell them the published Polyshares rule and stop. You are not their counsel. Polyshares publishes that client-held material is out of scope and that nothing leaves before an accepted offer. The micro1 terms I am using list excluded categories, but I could not re-check them publicly; get current terms from the buyer before relaying that list. The company and its lawyer decide if their contracts allow it.

If they ask who you are, say you are an independent referrer, name the one buyer you have in mind, and give the referral link. Do not sign as the buyer. Do not say you are micro1's partner.

If they ask how much, give the buyer-stated range you already used, repeat that their number comes from an offer, and do not invent a quote. Three people asked this. It is not the main objection. Do not let it take over the first reply.

If they ask for a call, book the call. On this file a booked call is the result we can count. Say that on the call too. A call is not a license.

Become the broker only if you still never hold the file

Three seats. A referrer introduces the company and the buyer does the rest. A sourcing agreement is a written deal with one buyer: you still do not hold the file, and you negotiate the fee, how long credit lasts after you stop, and what happens to unpaid fees if they end the agreement. A principal broker signs the seller and sells onward, which means someone on your side holds the data.

The brokerage plan I am using, dated 2026-10-11, says stay a referrer, and do not build the principal shop yet. The step it treats as worth taking is a written sourcing agreement with one buyer that does the rest of the work. It names Polyshares as that first buyer. That names a buyer to refer to. It does not make you their partner, and the micro1 terms I am using still bar you from calling yourself micro1's agent, recruiter, or partner. The company-facing Polyshares site says the company pays no fee. Ryan Lock, a Polyshares partner, posted on 2026-10-07 that they pay data brokers 6% on top of the price paid to the company. His example is $18,000 on a $300,000 price, plus 1% for referring another broker. That post is not a fee in our campaign file, and it is not a number to promise a founder.

Those same 2026-10-08 terms say that if no payout is specified, the referral payout equals 7.5% of the license or purchase price paid to the company. The referral page still says up to $50K per referral. Which of those two applies to a given deal is not published. Their sub-affiliate page offers a 20% bonus on a recruit's successful referral. Unpaid payouts are forfeited if the agreement ends. Do not promise a founder either number. Do not apply the 7.5% term or the $50K cap to a Polyshares license.

The four Polyshares amounts add to $1,135,000. Their average is $283,750. Those are buyer-stated licenses, not deals we closed. Polyshares' own site does not publish a percent to multiply them by. Use Lock's 6% only as what he posted, not as an expected check.

*The two referral paths, Polyshares then micro1. Deal sizes are buyer-stated. Polyshares' 6% comes from a partner's post; micro1 has not published which referral formula applies to a given deal. No payout is on file.*

Never hold the file. This is the default in that plan.

1. You email and qualify. You keep contact details only.
2. You ask headcount, tools, years of history, whether they own the record, and whether customer contracts restrict sharing. Answers only. No exports, no samples, no screenshots.
3. You make one introduction, or you pass one tracked link.
4. The seller connects their systems to the buyer.
5. You log introduced, offer made, signed, and paid.

The plan marks the added cost of that log, beyond outbound you already run, as near zero, plus a part-time deal owner. Both marks are estimates. There is no invoice in the note.

A screenshot or a sample changes the seat. You then hold personal data, including employee and customer detail that lives in chat and email. The plan sends the legal consequences to counsel: whether that is a sale of personal information, and whether you must register as a data broker. The California Privacy Protection Agency's own page sets the 2026 registration fee at $6,000. A law-firm report says the agency voted on 2026-08-06 and 2026-08-07 to raise that fee to $9,500 starting January 2027. Third-party writeups put Texas at $300 and the Vermont statute at $100, with a reported rise to $900 on 2027-01-01. Connecticut is reported at $2,500, and the registration duty is described as starting 2027-01-01 even though other parts of that law took effect 2026-10-01. This is not legal advice.

The same note estimates $250,000 to $400,000 to run a principal brokerage in year one, and it says every estimated line needs a real quote before anyone budgets it. The constraint it names is a buyer who will sign with you. That research did not find one signed for a principal shop.

The triggers it wrote down:

• Add other buyer routes after the first paid referral. It names Mercor and Troveo as later routes, and the micro1 20% sub-affiliate if another operator wants in.
• Ask for a written sourcing agreement after 5 paid deals, or $100,000 of referral income inside six months, or a buyer asking for more volume. Ask for a percent of the license price, credit for 12 months after you stop sending, pay when the buyer is paid, no forfeiture if the agreement ends, and the right to place a declined seller somewhere else. Counsel to read one agreement is estimated at $1,000 to $5,000.
• Go principal only when all of these are true: 15 or more closed deals in a year, a named buyer willing to sign a purchase agreement with you, staying a referrer is costing more than an estimated $300,000 a year in margin, and counsel has signed the contract set. Even then a processor holds the files and you hold the paper.

Customer contracts and old privacy promises can kill a deal even when you never touch the file. The plan also cites Slack's API terms, which bar an app from using Slack API data to train a large language model. That bar is not the same thing as a company licensing an export it already holds. Ask the pre-screen questions. Do not give the company rights advice.

Count the call, and leave the payout blank

Six of the 10 booked meetings came from the Google campaign of 5,078. Two came from the architecture campaign, where contacted volume is unknown. One came from the October trades campaign, where sends are unknown. One is unattributed. Attendance is blank for all 10.

Mail host is the result I would bet the next list on. 0 human replies out of 895 is a large zero, and a 39% bounce on the other-host list is a broken file.

Build the 5,128-firm trades list down to Google-hosted mail, confirm the job system before you name it, and send the reasons email with a buyer-stated range and no borrowed $898K. Suppress every domain already contacted in the campaigns above. One company, one buyer. When a person replies, answer what the data is. When they book, count a call. If someone asks what we got paid, no payout is on file.

Polyshares: https://polyshares.com/signup?invite=a4de836f166726d27c21f826d5da9b6e

micro1: https://refer.micro1.ai/register/?affiliateCode=f6ad03af-a739-43c1-bab1-5ac52039e45a&utm_source=referral&utm_medium=share&utm_campaign=job_referral