Article

Can you review the records before use?

By TrustPast

Updated October 24, 2026

Before a buyer uses fifteen years of your job files, you’ll want to read some of them as the buyer will. Not the originals. The prepared copy, with identifying details removed or transformed. This page settles whether you get that chance under micro1’s published terms, and how to run the review well: who reads, what they look for, what a cleaned estimate, change order and repair note look like, and what to do when something slipped through.

Other pages cover the ground around this one. Our page on how de-identification works shows what comes out of each kind of record. Our page on whether the buyer sees your customers’ names maps where names sit in your systems. Here we stay with one afternoon: the one when the samples arrive.

The short answer, in micro1’s words

Yes, you see samples first. micro1 says it twice on its data partnerships page. Once in its answer on how company data is protected: “companies have the opportunity to review representative samples before use”. And once in its answer on what happens after data is shared: “companies are given the opportunity to review representative samples before prepared data is used.”

Three words in that second sentence carry the weight. The word “representative” means the samples should look like the whole set. The word “prepared” means you see them after the identifying details are handled. The word “before” means nothing is used until you’ve had the chance to look.

The preparation itself is micro1’s job, not yours. Its page asks, “Does my company need to anonymize its data before transferring it?” The answer: “No, micro1 has processes for de-identifying and transforming sensitive information.”

The rules for that preparation are settled early. Under “Built with privacy and control first”, micro1 says: “Before any participation, together we confirm:” and lists four points. One is headed “De-identified before sharing”, with the line “Personal and identifying information is removed or transformed before your data is shared with third parties”. Another is “Defined retention periods”: “Raw data is deleted according to an agreed schedule”.

The scope is settled early too. micro1’s first step is “Discovery & pricing”, where it says: “We identify licensable datasets across your business and tailor an offer based on the nature, scale, and value of your data.” Its second step, “Contracting”, includes “a commitment not to resell or distribute your data until personal and identifying information has been anonymized.”

So the order runs like this, in our own words. The systems and the rules are agreed before anything moves. micro1 prepares the records. You read samples of the prepared copy. Then the records are used.

Apply & License Your Data

Two reviews, and only one of them is yours

Owners sometimes mix up two different checks. It helps to keep them apart.

The first check is micro1’s. Its third step, “Delivery & acceptance”, says: “You securely transfer your data through our APIs or other approved methods. We review the data for completeness.” Elsewhere it says that once a company transfers its data, “micro1 has 30 days to review and approve it.” That review asks whether the records are all there and usable. It is micro1 checking what it bought.

The second check is yours. The sample review asks a different question: did anything come through that shouldn’t have? A customer’s name in a technician’s note. A card number in a payment remark. A drawing that belongs to a client. Nobody else in the chain knows your files well enough to catch those.

Note the limit. micro1 does not publish how many samples you see, in what format, how long you have to read them, or what happens if one fails. It does not publish who sends them or when. Those are questions for the call. micro1’s referral page says qualified companies are invited to connect with a Data Partnership Manager “to discuss their data, licensing terms, and next steps.” That is the moment to ask.

What “representative” should mean for your files

A sample is only useful if it comes from the places where mistakes happen. Tidy structured fields rarely fail. The free text does.

As our own advice, ask that the samples cover four things:

Every system in the licence. If the job software, the shared inbox and the estimating tool are all in scope, you want records from each.
Every span of years. Software changes. A company that moved systems in 2017 has two kinds of files, and the older kind is usually messier.
Every record type. Estimates, change orders, service tickets, dispatch notes, inspection reports. Each has its own habits.
The worst fields. Technician notes, “misc” boxes, email bodies, scanned attachments. Ask for some of these on purpose.

Ask, too, that each sample carry a reference you can trace back. A job number or a record ID lets you open the original and compare. Without it, you can only guess what was removed.

How many is enough? micro1 does not say, and we won’t invent a figure for it. Our own rule of thumb is a habit, not a number. Keep reading until a whole batch from one source comes back clean.

Who in the company should read the samples

This section is our own advice. One reader is not enough, because no single person knows every kind of detail that matters. Three readers, each looking for one thing, work better than one reader looking for everything.

The person who answers your phones reads for names. Your dispatcher or office manager has heard every customer’s surname and street. They’ll recognise a regular account from a description of the job alone. Give them the samples first.

Your bookkeeper or controller reads for payments. They know what a payment remark looks like in your software. “Paid by check,” “card on file,” “financing approved,” “sent to collections.” They also know which notes mention a deposit or a lien.

A project manager or senior estimator reads for client-owned work. They know which jobs came with an engineer’s drawings, a property manager’s spec book or a national account’s service procedure. Those documents may belong to the client under your contract. Our page on whether licensing company records is legal explains why counsel reads those contracts first.

The owner signs off last. Not as a fourth reader, but as the person who hears what the three found and decides.

Keep the readers inside the company. Don’t forward the samples. Read them on a company machine, in the office, and keep one log.

Setting up the afternoon

The review goes faster with four things on the desk before anyone opens a file. These are our own suggestions.

A list of your largest accounts. Twenty commercial clients and your most frequent residential customers, by name and street.
Your staff roster. Your own people are named in job notes more often than customers are.
Your subcontractors and suppliers. A sub’s company name or a rep’s mobile number can sit in any change-order thread.
The contracts list from counsel. Which clients own their drawings, specs or reports, and which have confidentiality terms.

Then a log sheet with five columns: the sample’s reference, the system it came from, what was found, which reader found it, and whether it looks like a one-off or a pattern. That last column decides what you ask for next.

What to check: names, and the things that act like names

The obvious names usually come out cleanly. The slips are in the details that point to one job without naming anyone. As our own advice, read for these:

a first name left in a greeting, a sign-off or a “per” line, as in “per [first name], leave the old valve”;
names of tenants, inspectors, building engineers and the client’s own staff, not just the customer;
permit numbers, which a city’s public lookup can turn back into an address;
purchase order numbers and a commercial client’s internal job codes;
a street name or landmark left in the scope text, such as “the third house past the water tower”;
photo captions and file names that still carry a surname.

Your readers know which of these your company writes most. A dispatcher who types cross streets into every ticket will find them faster than anyone.

What to check: payments

As our own reading, customer names and payments are never what is wanted. Your bookkeeper is reading to confirm that.

A price and a payment are different things. A line price on an estimate can be part of the reasoning, and micro1 says datasets may undergo “synthetic rewrites that preserve workflow structure while reducing direct ties to original records.” As our own reading, a rewritten figure can keep the logic of a price while losing its tie to one job. Whether prices stay, change or go is something to settle on the call.

A payment is never part of how the work was done. As our own advice, flag any of these that survive preparation: card digits, check numbers, bank names, deposit receipts, financing approvals, collection notes and any remark about whether a customer paid on time.

What to check: work a client owns

This is the check owners forget, and the one a name search won’t catch.

Some files in a job folder were never yours. A mechanical engineer’s stamped drawings. A general contractor’s schedule. A property manager’s maintenance spec. A national account’s own procedure for its stores. They sit in your records because you needed them to do the job.

Your project manager reads the samples with counsel’s contracts list beside them. If a sample shows a client’s document, or a long passage copied from one, it goes in the log. As our own advice, ask for that source to come out of scope altogether rather than be cleaned record by record. Our page on what is never included in a records licence has more on client-owned work.

Three samples, read the way a reviewer reads them

Each sample below is our own illustration, written for this page. None is a real record, and none is micro1’s. Each one shows what passes and one thing a reader would flag.

An estimate. A commercial electrical job.

Estimate rev 3, service upgrade, commercial client, multi-tenant retail building. Rev 1 priced a 400-amp panel swap. Utility site check found the transformer undersized for the added load; rev 2 adds a utility coordination fee and a 6-week lead time. Rev 3 splits the work into two nights to keep the anchor tenant open. Permit [number left in]. Client chose rev 3.

What passes: the revisions, the utility finding, the reason for the two-night split. That is an estimate becoming a job. What gets flagged: the permit number. Your reader logs it, and asks whether permit numbers were in the redaction rule at all.

A change order. An email thread on a tenant fit-out.

From: Project manager. To: Owner’s representative. Subject: Change order 7. Ceiling opened at gridline C. Found abandoned ductwork blocking the new sprinkler route. Removing and capping adds one day and [amount].

Reply: Wasn’t that on the survey?

Reply: The survey showed the live duct only. We’ll absorb the capping, you cover removal. Thanks, [first name left in]

What passes: the hidden condition, the price, the argument and the split. That is a change order being settled. What gets flagged: the first name in the sign-off. Sign-offs are a pattern, not a one-off, so the log says pattern. Our page on email archives covers why the threads are worth keeping.

A repair diagnosis. A service ticket from a commercial kitchen.

Service ticket. Restaurant client. Complaint: floor drains backing up during dinner service, third call. Tech ran camera from cleanout: grease build-up and a belly in the line 30 feet out. Jetted line as short-term fix; recommended excavation of the sag and a larger interceptor. Paid on card ending [four digits left in].

What passes: the camera run, the finding, the short-term fix and the recommendation. That is a repair being diagnosed. What gets flagged: the card digits. A payment remark like that should never survive. Our page on job files and work orders explains what else a ticket like this holds.

Notice what none of the flags touched. The work stayed whole each time. The fix is to remove a detail, not to lose a record.

What to do when something slipped through

Expect the first batch to turn up something. If it turns up nothing, read more records before you sign. This section is our own advice, because micro1 does not publish what follows a failed sample.

Sort the find first. A one-off is a single stray detail in one record. A pattern is something the rule missed everywhere, like every sign-off or every permit number. A wrong source is a whole system or folder that shouldn’t be in scope, like client drawings or the payments log.

Report it in writing. Send the log entry: the reference, the field, what was found. Don’t paste the original detail into the email. The reference is enough for micro1 to find it.

Ask for the rule to change, not the record. A one-off fixed by hand leaves its cousins in the other ten thousand files. Ask that the fix apply to every record from that source.

Ask for a fresh sample. Same system, same years, after the fix. Read it the same way.

Pull the source if it keeps failing. If a folder keeps coming back with names or client documents, ask for it to come out of scope. A smaller licence that is clean beats a larger one you worry about.

Don’t edit your own systems. Your files stay as they are. micro1 prepares the copy, and micro1’s answer on anonymizing first is that your company doesn’t need to.

Two more questions to ask on the call, before you ever see a sample. What happens to the pace of the deal if a batch fails? And does a narrower scope change the offer? micro1 says the final payout “is determined based on data sources, volume, quality, access terms, and due diligence.” Better to hear the answer early.

Signing off

When a batch comes back clean from every source, the owner signs. As our own advice, make the sign-off a short written note, not a nod on a call. It names the systems reviewed, the years covered, the number of samples read, what was found and how each find was resolved. Keep a copy with the agreement.

That note is useful later. If a partner, a lender or a large client ever asks how your records were handled, it answers the question in one page.

Why this review is shorter than owners expect

Owners picture reading every file. They don’t have to, and the reason is what the buyer wants.

micro1 says: “We are looking for data that reflects how work is actually performed within an organization.” It adds: “The most valuable datasets capture real operational expertise, decision-making processes, and repeatable workflows”. Its 9 October 2026 announcement says micro1 “uses de-identified company data”.

So the records are de-identified before use, and the review checks that. Customer names and payments are never what is wanted. What is wanted is how the work was done. How an estimate became a job. How a change order was settled. How a repair was diagnosed. Your readers are not judging the work. They’re only looking for the few details the work never needed.

Our cornerstone page on what data buyers want from a trade business goes through that split trade by trade. A drain line doesn’t care whose restaurant it runs under.

What this means for a company with 30 or more staff

Here is one company, as our own hypothetical and not a description from micro1. Say a commercial plumbing and mechanical contractor with 60 people and fourteen years in its job software. micro1’s referral page lists “30+ employees” among its company requirements, so it is in range. Our page on companies with 30 or more staff covers that lane.

The records it keeps. Service tickets with technician notes. Work orders and job files for tenant fit-outs. Estimates with every revision. A shared inbox full of change-order threads with general contractors. Inspection reports and backflow test records.

Who reads the samples. The dispatch lead reads for names, with the top thirty accounts beside her. The controller reads for payments and lien notes. The senior project manager reads for engineers’ drawings and the national accounts’ procedures. The owner signs.

What they would ask for up front. Samples from the old system before 2016 and from the current one. Samples from the shared inbox, not just the tickets. A reference on every sample. A fresh batch after any fix.

What should survive. How a backflow failure was traced to a fouled check valve. How a fit-out change order was split with the general contractor. How a repeat drain call ended in an excavation recommendation. That is the record a buyer licenses to train AI models on real work.

Worth two minutes to see your range? Our estimate asks five questions.

What is never included

micro1’s own words on this are short. Its data partnerships page says: “We protect sensitive information through data anonymization and secure handling.” Four points, including de-identification and retention, are confirmed with each company before any participation. It does not publish a fixed list of what stays out.

Our own list goes further than micro1’s line, and it is ours, not micro1’s. Unless micro1 agrees otherwise in writing, the following stay out of the licence: attorney-client privileged records, health information, payment-card and financial-account details, Social Security numbers, minors’ data, biometric data, and classified or export-controlled material. Two more belong on a review page. The customer list itself. And anything a client owns under its contract with you.

The sample review is the place to confirm that none of these came through. If one did, it is a wrong source, not a one-off. Our How it’s safe page gives the short version, and our page on whether it’s safe to license company records covers access, retention and deletion.

Where we fit

We, TrustPast, explain how this works and give you a range with our estimate. The range is ours, not an offer. You apply to micro1 yourself, through the button below. micro1 pays us a referral fee. You pay us nothing.

We never see your files, and we never see the samples. The review is between your company and micro1. This page is only a way to walk into that afternoon with the right people at the desk and a log sheet ready.

TrustPast

Questions and answers

Do we get to see our records before micro1 uses them?

Yes, as samples. micro1 says “companies are given the opportunity to review representative samples before prepared data is used.” The samples are of the prepared data, so you see the records after identifying details have been removed or transformed.

Do we have to take the names out ourselves first?

No. micro1 answers that question in one line: “No, micro1 has processes for de-identifying and transforming sensitive information.” Your job is to read the samples and say what still needs to come out.

How many samples will we see, and how long do we have?

micro1 does not publish a number or a deadline. As our own advice, ask both on the call with the Data Partnership Manager, and ask that the samples cover every system and every span of years in the licence.

What if a customer’s name or a card number is still in a sample?

micro1 does not publish what happens after a sample shows a slip. As our own advice, write it down with the record and the field, report it in writing, ask that the rule be fixed across the whole set, and ask for a fresh sample from the same source before anything is used.

Who in our company should read the samples?

This is our own advice, not micro1’s. The person who answers your phones, for customer names. Your bookkeeper, for payments. A project manager or estimator, for work a client owns. The owner signs off last.