10 February 20267 min read
Every buyer prices an outsourcing partner against a marketplace at some point, usually while looking at two rates that are not comparable. The marketplace number is lower and it is a real number. What it does not include is most of the work, and the work it does not include is the part that fails.
This is not an argument that marketplaces are bad. They are very good at a specific thing, and for that thing a partner is overhead you should not pay for. The mistake is using one where the other belongs.
What a marketplace is genuinely good at
Marketplaces are excellent when the work is bounded, the skill is narrow, and you can evaluate the result yourself. A logo. A one-off script. A landing page. A specific library integration where you can read the pull request. Work with a clear beginning and end, where the cost of a bad outcome is that you paid for something you throw away.
In those cases the marketplace's rate advantage is entirely real, because you are not buying continuity, coverage, or accountability, and a partner would be charging you for all three. Buying a partner for a two-week task is like hiring a recruitment firm to fill a summer internship.
What you are actually buying from a partner
A partner's margin covers work that otherwise lands on you. It is worth being explicit about what that work is, because if you do not value it, you should not pay for it.
- Pre-vetting: someone has already interviewed a large number of people so that the shortlist you see is short. This is the largest hidden cost of the marketplace route and the one buyers consistently underestimate.
- Replacement: when someone leaves, breaks a leg, or turns out not to be as described, it is the partner's problem and their bench, not your recruiting problem starting from zero.
- Continuity of knowledge: a team that keeps context across people, so the second engineer is not starting where the first one did.
- Coverage: holidays, illness, and the fact that one freelancer is a single point of failure for whatever they touched.
- The commercial and legal layer: one contract, one invoice, IP assignment, a data processing agreement, and a counterparty that exists as a company if something goes wrong.
- Management, in managed delivery: someone whose job is running the work, which is otherwise your job.
If your answer to most of those is that you can handle it internally, the marketplace is the rational choice and you should take the rate. Just make the decision knowingly rather than by comparing two numbers that measure different things.
The failure modes are different
Marketplace failure is usually fast and cheap. You hire someone, it does not work, you stop, you are out a few weeks. The exceptions are ugly: a freelancer who is the only person who understands a system, disappears, and leaves you with code you cannot maintain. Or the one who is quietly subcontracting the work onwards, which you find out about when the quality changes mid-project.
Partner failure is slower and more expensive. You do not usually get a bad engineer, you get an average one on a long engagement, and average is much harder to act on than bad. Or the account manager who was excellent in the sales process is invisible by month two. Or the good people are rotated onto a newer, larger account and you get their replacements without being told.
The mitigations are correspondingly different. Against the marketplace, insist that nothing has exactly one owner, and read the code. Against a partner, ask who is assigned, whether they are dedicated or shared, and how much notice you get before a change, and put the answer in the contract.
The vetting problem, which is where the time goes
The rate comparison ignores what it costs you to find the person. On a marketplace you are the filter, and filtering is not a quick job: writing the posting, reading a large number of applications of which most are automated, running screens, running a technical exercise, and doing it again when the first choice does not work out.
That cost is paid in senior internal time, which is the scarcest thing you have and the reason you were outsourcing in the first place. It is also paid repeatedly, because freelancers move on. A partner amortises it: the vetting happens once, for many buyers, by people who do it constantly.
Contracts, IP and data
For a European company this is where the routes separate most sharply, and it has nothing to do with quality of work. With a partner you sign one contract with a company: IP assignment, confidentiality, a data processing agreement, defined subprocessors, and a legal entity that continues to exist. With a marketplace you have a platform's standard terms and a person, whose IP assignment may depend on the platform's boilerplate and whose jurisdiction may make enforcement theoretical.
If the work touches personal data, this stops being paperwork. You remain the controller regardless, so the processing arrangement has to be real. Plenty of freelance arrangements can be made compliant with a direct contract, and many companies do exactly that. It is just work, it is legal work, and it belongs in the comparison rather than being discovered later.
How to choose, in one pass
Use the marketplace when the work is bounded, you can judge the output yourself, the loss if it fails is a few weeks, and nothing you cannot rebuild depends on one person's continued attention.
Use a partner when the work continues, when continuity matters more than the rate, when you need coverage across a working day rather than one person's availability, when you do not have senior time to spend on screening, or when the contractual position has to hold up to your own legal and security review.
And note that the honest version of the second list is mostly about your constraints rather than the provider's virtues. That is the right way round. What you are buying from a partner is the removal of specific work from your plate, and you should be able to name which work, item by item, before you agree to pay for it.