Ask a freelancer why they stay on a platform they complain about, and the answer is rarely the tooling or the fees. It is the reviews. Four years of work, a hundred and forty completed jobs, a rating that took a decade of professional life to accumulate, and none of it can leave.
That is the real lock-in. Fees are visible, so people can compare them. Reputation is harder to see as a cost, and harder to move. That makes it the most effective retention mechanism any marketplace has. It also means nobody has to defend it, because nobody frames it as a policy.
The real lock-in is the history a worker cannot move
A marketplace track record looks like one thing from the outside. When you break it apart, it is made from several kinds of evidence, and those kinds behave differently.
The record usually contains these parts:
- Verified facts. Jobs completed, value transacted, on-time rate, dispute count. Objective, machine-checkable, and generated by the platform’s own systems.
- Client testimony. Ratings and written reviews. Subjective, but attributable to a specific person who can be contacted.
- Behavioural signals. Response time, rehire rate, cancellation rate. Derived, and the most useful predictor of future behaviour.
- Identity and credentials. Who someone is, what they are qualified to do, whether they are insured. Portable, and the piece most platforms already accept from third parties.
The second category is the difficult one to move, because a review written on one platform is testimony about a job that happened there. The other categories are either facts or are derived from facts. There is no principled reason they cannot travel.
Platforms have real reasons to protect reputation data
Platforms have a real case, and it deserves to be stated properly.
A review only means something in a context where it can be verified. If reputation is portable, it becomes forgeable. A new platform accepts imported ratings, bad actors import fabricated ones, and the signal degrades for everyone. The lock is what makes the data trustworthy.
There is also a fairness argument. The platform funded the systems that generated the record, absorbed the fraud risk, and adjudicated the disputes that make the numbers meaningful. Letting a competitor free-ride on that is not obviously reasonable.
Both points are legitimate. They justify verification and attribution. They do not justify immobility. You can solve the forgery problem with signatures rather than with walls.
A portable record has to prove where it came from
Portability is not a CSV export. Everyone has a CSV export, and it is worthless, because a spreadsheet of self-reported ratings proves nothing.
Portable means a receiving platform can verify the claim without trusting the person carrying it. That requires three properties.
- Signed by the issuer. The record is cryptographically signed by the platform that generated it. Anyone can check the signature. Nobody can fabricate a record without the private key.
- Attributed, not laundered. The receiving platform displays it as what it is: forty jobs verified on another named platform, not forty jobs here. The origin travels with the claim, and users can weight it themselves.
- Revocable. If a record is later invalidated, fraud discovered, a review removed, the issuer can publish a revocation that consumers of the record can check. Without this, portability exports mistakes permanently.
None of that is novel. It is how academic credentials, professional certifications and increasingly employment history already work.
A useful export needs the actual fields
Vague talk about portability is easy. The real question is what we sign and export.
A useful record contains these fields:
- Issuer and subject. Which platform generated the record, which account it describes, and when it was produced. Records expire, because a two-year-old snapshot is not evidence of current behaviour.
- Volume. Engagements completed, total value transacted in a stated currency, and the date range they span. Value bands rather than exact figures, because a precise number is commercially sensitive and a band is enough to establish scale.
- Reliability. On-time completion rate, cancellation rate initiated by each side, and average time to first response. These are the signals that actually predict how someone will behave, and they are computed rather than claimed.
- Disputes. Count raised, count resolved in the subject’s favour, count against. Exporting only the favourable half would make the record worthless, so the full picture travels or nothing does.
- Testimony. Individual reviews with the reviewer’s verified status, the date, the engagement value band and the text. Attributed to a real completed engagement rather than aggregated into a score.
- Identity assertions. Which checks were performed, by whom, and when. Right to work, insurance, professional registration. These are the most portable elements because they were never really about our platform in the first place.
Forgery is handled with signatures and revocation
The objection that portability enables fraud is the strongest one. It also has a well-understood answer that predates marketplaces entirely.
The record is signed with a key held by the issuing platform. A receiving platform fetches the issuer’s public key from a published location and verifies the signature. A forged record fails verification. A modified record fails verification. A record from a platform nobody has heard of verifies correctly and is displayed with that platform’s name attached, letting the viewer decide what it is worth.
Revocation is the part people forget. If a review is removed after a fraud investigation, the exported record describing it is still floating around signed and valid. So the issuer publishes a revocation list, and any consumer checks it before displaying. Without that step, portability makes mistakes permanent, which is worse than not having it.
All of this is ordinary public-key infrastructure: keys, signatures, verification and revocation. The reason it does not exist in this market is not that it is hard. The parties who could deploy it have no reason to.
Open Lance exports facts and treats reviews as testimony
On Open Lance the reputation record is exportable as a signed document containing the verified facts and the behavioural signals, with reviews included as attributed testimony rather than as a portable score.
That distinction matters. We export that a named client gave a four-star rating with this text on this date, signed by us. We do not export an aggregate rating for another platform to display as its own, because an average computed under our rules means nothing under theirs.
We also accept imported records, displayed as what they are, with the issuing platform named and the signature checked. A new arrival with a verified history elsewhere is not treated as a brand-new account, and is not treated as an established one either.
Imported records should shorten the cold start
Importing a record is not the same as trusting it. Most of the design sits in that difference.
An imported history should shorten the cold start, not eliminate it. A new arrival with four years of verified work elsewhere should skip the worst of the empty-profile problem.
That verified history can affect three things:
- Appear in search results rather than being buried.
- Be eligible for engagement sizes a brand-new account would not be.
- Avoid the worst of the empty-profile problem.
They should not inherit a top rating on a platform where they have completed nothing.
The honest presentation is two separate figures side by side: what has happened here, and what is verified elsewhere. Merging them into one number is the failure mode, because it launders an unverifiable average into an apparently native one and destroys the meaning of both.
We also weight imported records by issuer over time. A platform whose imported users go on to perform well here is worth more. One whose users disappoint is worth less. That is an ordinary reputation problem applied one level up. It is how a network of portable records eventually polices itself without a central authority deciding who is legitimate.
Portability forces the platform to earn retention
This makes it easier for our own users to leave. That is deliberate, and the argument rests on it. It is also not altruism.
A platform whose retention depends on trapping reputation has a structural incentive to under-invest in everything else, because the users cannot go anywhere regardless. Remove the trap and the platform has to be worth using, continuously, which is a healthier thing to be subject to.
It is also commercially rational for us specifically. We are not the incumbent. Portability threatens whoever holds the largest hostage population, and that is not us. Arguing for it costs us little and costs them a great deal. We should be honest about that, rather than pretending the position is purely principled.
This has been blocked by incentives, not engineering
There are three reasons this has not happened already, and none of them are technical.
- No incumbent benefits. Portability is worth most to the platform with the fewest users. The ones with standing to set a standard have every reason not to.
- There is no neutral registry. Signatures need a way to check which keys belong to which platform. That is a coordination problem requiring an institution nobody has built, because nobody profitable wants it.
- Users do not demand it until they leave. Reputation lock-in is invisible while you are staying. It becomes obvious exactly once, at the moment it is too late to have wanted the feature.
Freelancers can protect some of the record now
Waiting for platforms to agree on a standard is not a plan. Some practical steps work today.
- Keep your own record. A simple document listing engagements, dates, scope and outcome, maintained as you go rather than reconstructed later. It is not verifiable, but it is the raw material for everything else.
- Ask clients for testimony you own. A short written reference, given directly, is portable by definition. Most clients will provide one and almost nobody asks.
- Keep evidence of the work itself where you are permitted to. A portfolio is reputation you control.
- Treat the export question as a pricing question when choosing a platform. A platform that will not let you leave with your record is charging you a switching cost that does not appear in its fee schedule.
Large buyers are likely to make portability happen
The employment market has been moving toward verifiable credentials for a decade, driven by regulators and large employers who got tired of unverifiable CVs. Freelance marketplaces have avoided the pressure because their users are individuals with no bargaining power.
That changes when a sufficiently large buyer wants it. An enterprise engaging hundreds of contractors across several platforms has a genuine problem: it cannot compare them, and it cannot carry a vetted supplier from one to another. That buyer has leverage no individual freelancer has.
Our expectation is that portability arrives through procurement rather than through platform goodwill, and that the platforms which support it early will be the ones enterprises can actually use. We built it now because it is cheap to build before you have millions of users and expensive afterwards, and because arguing for it while it costs us nothing is the only time the argument is worth anything.
The reasoning connects directly to why we charge no commission: both are bets that competing on lock-in is a weaker position than competing on being worth staying with. The full economics are in the Open Lance case study.
