A Care.com product manager opened the company's mobile cancellation screen during a test and typed three words into an internal document: "Dark UX Pattern." That label told the whole story before the FTC filed its complaint. The platform sold access to 4.7 million job postings while more than half of them came from accounts that could never hire anyone.
Every operator who has paid for access to a lead list, a job board, or a vendor platform has heard the same pitch. The number comes first. "We have X thousand listings." "Our network reaches Y million users." The number is supposed to prove the market works. It does not prove that. It proves someone counted.
What the FTC Found
The research on this one is worth reading in full. The FTC's complaint against Care.com, which led to an $8.5 million settlement and refunds to 194,207 people, laid out a clean anatomy of phantom supply.
Between 2019 and early 2022, Care carried 4.7 million job postings. Fifty-six percent of those postings came from free accounts. Free members could not see who applied to their listings. They could not read messages from job seekers. They could not hire through the platform. The postings were decorative. They sat in the search results and made the market look full.
Care sold 2.9 million paid memberships during that same window. Every one of those members paid for access to a supply pool that was more than half phantom. The number was not a side effect of growth. The number was the product.
Then the earnings layer. Care told job seekers they could earn $18 an hour through the platform. That figure ran on third-party job sites. On Care's own platform, the listed rates for the same roles ran between $13 and $14.50 an hour. The gap was not a rounding error. It was a second manufactured number, pointed outward to pull people in.
The FTC found that Care never tracked what caregivers actually earned through jobs found on the site. The platform sold an earnings figure it had no data to support. It kept selling that figure after the FTC sent a formal penalty notice in 2021. They knew. They kept going. The big number does not fail people. It fails the system people are trying to run.
The Structural Flaw
Care's failure was not a customer service problem or a pricing mistake. The flaw sits one level deeper. When a platform treats its supply count as the thing it sells, the count becomes the product. The conversion that supply should produce stops being the point.
The problem is not that the number was wrong. The problem is that the number replaced the one that mattered: how many of those listings could lead to a completed transaction.
The Pattern at Scale
This is not one company's mistake. Amazon was sued by the FTC for $2.5 billion over a subscription system that employees called the "Iliad Flow" in internal documents. The name came from Homer's epic, a nod to how long the path ran: four pages, six clicks, fifteen options before a subscriber could leave. Same architecture. Same principle. The friction was the feature, and the subscriber count was the number the company sold to the market.
I have made this mistake. Most operators in a buying position have made this mistake. You see a vendor's listed supply, and you treat it as proof of a working market. It is not proof. It is a count. The two are not the same thing.
Three Moves Before You Pay
Once that is clear, three moves follow from it.
Move 1: Ask for the Conversion Rate, Not the Listing Count
Before you pay for access to any platform, job board, or lead database, ask one question. Of the listed supply, what share resulted in a completed transaction in the past 90 days? If the vendor cannot answer, or answers with a user count instead, the supply number is decorative. That question costs nothing to ask. It costs real money to skip.
Move 2: Test a Small Batch Before You Buy the Full Pool
Pay for the smallest tier on offer. Run ten leads, ten listings, ten contacts through your own pipeline. Track how many respond, how many convert, and how many were reachable at all. The ratio you get from ten tells you more than the vendor's number for ten thousand.
Move 3: Compare the External Claim to the On-Platform Reality
Look at what the vendor tells the outside world versus what shows up once you are inside the system. Care ran $18 an hour on job sites. The platform showed $13 to $14.50. If the number the vendor uses to sell you access does not match the number you see once you are in, the gap is the tell.
What the Audit Shows
Running these three moves for 90 days does something the vendor's pitch never did:
Which supply numbers are tied to real conversion and which are just volume.
Whether the vendor's external claims hold up inside the actual system.
Where the friction lives, in the platform's design or in the market itself.
What your real cost per converted lead is, not the one in the sales deck.
The Feedback Loop
At the end of 90 days, ask three things.
→ What moved the number? Which leads, listings, or contacts led to a real outcome?
→ What looked like progress but left no trace? High response counts with no closed deals. Full inboxes with no revenue.
→ What friction showed up more than once? The same dead-end listing type. The same gap between the promised number and the real one.
That is the difference between a vendor pitch that sounds right and a system that proves itself.
Where You Stand
A product manager at Care.com typed three words into an internal document and described the whole model in a sentence. The FTC took years to reach the same conclusion. Your version of that document is your own conversion data. If you are paying for access to a number you have never audited, the number is not serving your business. It is the business you are paying for.
