Construction workers face real harassment and nowhere safe to report it. Does this idea actually solve that?
There's a moment in this idea that genuinely got me. A worker on a 40-person jobsite experiences a racial slur from a foreman. He knows HR is the foreman's drinking buddy. He knows the QR code on the trailer wall goes to an email inbox managed by the same office manager who processes his paychecks. He says nothing. The job ends in six weeks anyway.
That moment is real. EEOC data backs it up: construction is consistently in the top five industries for harassment charges filed, year after year. Reddit threads about jobsite harassment rack up hundreds of upvotes and comment sections full of workers describing exactly the fear-of-retaliation dynamic this idea is trying to solve. The problem is not imaginary.
The existing tools genuinely don't cover this. Procore is a project management platform that bolted on safety incident reporting. SafetyCulture (iAuditor) is a checklist and inspection tool. Neither was built to handle a worker saying "my supervisor threatened me yesterday" with any kind of legal chain-of-custody or anonymity protection. NAVEX and Vault Platform do exist for enterprise ethics reporting, but they're priced and built for Fortune 500 legal and compliance teams, not a 60-person roofing subcontractor in Dallas.
The QR code insight is real too. No app install, no account creation, no login that ties a device to an identity. You scan, you report, you leave. For a transient workforce that cycles through jobs every 6-18 months and has deep reason to distrust anything the company sets up, that frictionless entry point is the right instinct. It's one of those ideas where the person proposing it clearly spent time thinking about the actual user, not just the buyer.
The business model math is defensible on paper. 80,000 addressable US firms with actual HR staff, $79/month average, 88% gross margins. You don't need much penetration to build something real. And the retention hook is genuinely good: once a firm has two years of incident history and audit logs in the platform, they're not migrating anywhere. That data becomes part of their EEOC insurance renewal workflow. That's real lock-in.
Here's the problem I can't get past: the product only works if workers actually use it. And workers will only use it if they genuinely believe it's anonymous. On a 40-person jobsite, true anonymity is nearly impossible to guarantee, and the workers who most need to report something are exactly the ones smart enough to know this.
Think about it from the worker's perspective. Five people witnessed the incident. Three of them wouldn't report it. The other one already left for another job. That leaves you. You fill out the anonymous form. The HR admin receives a report about an incident that happened last Tuesday at 2pm on the east scaffold. There were two witnesses named. Only one person fits that description. Your anonymity lasted about four seconds.
No UX fix solves this. It's a structural problem with small-firm anonymous reporting, and it's the same reason most ethics hotlines at small companies are effectively decoration. The product is most useful at firms large enough that individual reports can genuinely be de-identified, which pushes you toward 150+ employee firms, which is a much smaller slice of the market than the pitch implies.
The buyer persona problem is the second thing that worries me. The idea targets "HR managers at small-to-mid sized GC firms." But below about 50 employees, there is no HR manager. There's Karen in the office who handles payroll, permits, accounts payable, and occasionally reminds people that the safety posters need to be updated. Karen has no software budget, no authority to approve a new SaaS subscription without the owner's sign-off, and when a harassment complaint lands on her desk, her first call is to the company's construction attorney, not to open a new dashboard.
You're really selling to firms with 50-150 employees, which narrows the addressable market considerably.
The regulatory exposure here is something I'd lose sleep over. If a state AG or plaintiff's attorney subpoenas the platform's records during an EEOC investigation or discrimination lawsuit, they can compel production of all submissions, including the ones the admin marked "resolved" or archived. That includes metadata. That metadata can identify reporters in small firms. The vendor could be the mechanism by which an employee's identity gets exposed in a legal proceeding, which destroys the product's entire value proposition in one news cycle.
This isn't hypothetical. NLRB investigations routinely subpoena employer HR records. The platform needs a lawyer who specializes in this before it signs its first customer, not after.
The union exposure is the other one. Construction is heavily unionized in most major metros. Union contracts have specific grievance procedures that are legally protected. If a union rep argues that this platform is being used to route complaints away from those protected channels, you've created labor relations liability for your customer and potentially for yourself. Your customer's attorney will hear about this and advise them to cancel before the pilot ends.
The competitive analysis notes that Procore has 45% market share and a sales force already calling on every target customer. The counterargument is that firms won't trust a Procore bolt-on for sensitive HR cases the same way they'd trust a purpose-built tool. I think that's partially true. I also think Procore doesn't need to build this themselves. They can acquire Vault Platform or a similar company for $5-10M and bundle it into their enterprise tier within 18 months. At that point, your differentiation story gets much harder to tell.
The other thing worth saying plainly: the "why now" argument, which is mobile adoption on sites and growing EEOC risk awareness, has been true for at least five years. The absence of a funded competitor in this niche is probably not a sign of an overlooked opportunity. It's more likely a sign that multiple teams have tried to crack this distribution problem and found it harder than expected.
Fear-based B2B sales to SMBs is brutal. Most 60-person GC owners have never faced an EEOC claim and have mentally discounted the probability to near zero. Selling them a tool to prevent something they don't believe will happen to them requires so much education before conversion that your CAC math starts looking very different from the model.
I want this to exist. The problem is real. The people being harmed on jobsites deserve better than "file a complaint with the same HR office that employs the person harassing you."
But I think the version described here, targeting small GCs with a reporter-side anonymous model, is genuinely vulnerable. Not fatally broken, but vulnerable enough that I'd want to change the strategy significantly before building.
If I were going to pursue this, I'd do two things differently. First, I'd move upmarket faster than the pitch suggests. The 50-150 employee range isn't the sweet spot. The 150-500 employee range is where there's a real HR professional with budget authority and enough workforce size that anonymity is actually achievable. That's a smaller number of firms, but the unit economics get better and the buyer persona is real.
Second, I'd think hard about whether the real product is the insurance and bonding angle. Construction firms pay a lot for general liability and workers' comp. If you can show a carrier that firms using this platform have lower harassment-related claims, you have a distribution partnership that doesn't require cold LinkedIn DMs to skeptical HR managers. You have an insurer telling their customers to use the platform as a condition of coverage. That changes everything about the distribution difficulty.
The validation test proposed, a Typeform and Notion pilot with three GC firms, is the right call. Do that before writing a line of production code. The specific success metric (two of three firms submit at least one real report within 30 days) is honest and concrete. I'd add one more condition: at least one worker submits a report without being prompted or reminded by management. That's the real test of whether workers trust it enough to use it voluntarily.
Don't build the full stack until you've passed that test. The QR code insight might be right. The market might be there. But the trust problem is real, and no amount of good engineering fixes a product that the people who most need it are afraid to touch.