Fixed price or hourly
Most launch jobs get routed by habit. You buy a packaged order because that is the shape you saw first, or you hire by the hour because the job felt too vague to price — and then you spend the engagement fighting the contract instead of the work. The routing decision has one test behind it, it is answerable in about a minute, and it does not require you to know what the job should cost.
The test
Can you enumerate the deliverable before any work starts?
Not estimate it. Enumerate it: a countable list of artefacts, in named formats, at stated quantities, where you and the seller would produce the same list independently. Eight product images, JPEG and WebP, on pure white, plus one lifestyle composite per product. That is enumerable. "Fix the product photography" is not.
If you can write that list, the job is closed and belongs on a fixed price. If you cannot write it — because what needs doing is part of what you are paying to find out — the job is open, and pricing it as though it were closed makes the number a fiction that someone has to absorb.
That is the whole decision. Everything below is either how to apply it when the answer is not obvious, or how to convert an open job into a closed one, which is usually the cheaper move.
Why the platform is the second question, not the first
The two marketplaces most launch buyers end up on are shaped around different default answers to that test, which is why the routing decision often looks like a platform decision.
Fiverr is built around packaged fixed-price orders: a seller publishes a defined scope at a defined price, and you buy it. That structure assumes the scope was closed before you arrived, which is why it fits enumerable jobs cleanly and fights you when the job is exploratory. Upwork supports both hourly contracts and fixed-price ones, so it can carry an open scope — you are hiring capacity and direction rather than buying a listed artefact.
Take that as a description of default shape, not a ranking, and check the current mechanics yourself rather than trusting this paragraph six months from now: contract types, fee structures, milestone and time-tracking behaviour, and dispute processes belong to the platforms and they change them. What does not change is the underlying logic — an enumerable deliverable priced per unit, or unenumerable work priced per hour with a review rhythm attached.
The practical consequence is that most of the launch tasks on this site are enumerable, so most of them route to a fixed price regardless of where you buy them. The interesting cases are the few that do not.
Where fixed price is the right shape
Fixed price works when the artefact is the point and the path to it is the seller's problem. You are buying an outcome you can count on delivery, so the seller absorbs their own inefficiency and you absorb nothing you did not agree to.
Route to fixed price when the deliverable list is countable and stable; the job is a category a seller has done many times before; you can check the delivery against your own list without watching the work happen; and a second attempt elsewhere would cost you time rather than a rebuild. Product images, a logo file set, copy for a known page structure, an ad cut from supplied footage, a defined email sequence — these are all closed jobs, and each has a brief on this site precisely because writing the brief is what closes them.
The failure mode is specific and worth naming, because it is nearly always the buyer's doing. A fixed price against a vague brief does not produce a fair negotiation about scope; it produces a delivery that satisfies the brief you actually wrote. The seller executes what they were given, you receive something defensible and wrong, and the revision rounds get spent absorbing your new thinking rather than correcting their work — so the real defects ship. That is scope creep running in your direction, and no contract shape protects you from it.
Where hourly is the right shape
Hourly works when the work is diagnostic, when the direction will change based on what is found, or when you genuinely need someone available over a period rather than a parcel delivered at the end of one.
Route to hourly when the first task is to find out what the tasks are; when you will be making decisions mid-job that change what remains; when the work is maintenance or on-call rather than a project; or when the deliverable is judgement — a review, an audit, a second opinion on an architecture — rather than an artefact.
Concretely, at launch: a theme behaving unpredictably across devices for reasons nobody has identified yet, an unexplained checkout drop, a store you inherited and want assessed before you decide what to commission. In each case the enumerable version does not exist yet, and demanding one just moves the guessing onto the seller, who will price the guess with a margin for being wrong.
Hourly has its own failure mode, and it is not sellers padding time. It is the absence of a review rhythm. Without a checkpoint, hourly turns into an open tab you notice late, and the conversation you eventually have is about the total rather than about the work. So set the rhythm when you open the contract: a cap for the first block of hours, a written summary at the end of it, and an explicit decision — continue, redirect, or stop — before more time is booked. That is not distrust, it is the thing that makes the shape work.
Closing an open scope is usually the cheaper move
Most jobs that feel open at launch are not genuinely exploratory. They are enumerable jobs you have not finished thinking about, and one hour of your own attention converts them.
Try this before you conclude a job needs hourly. Write the deliverable list as far as you can. Where you stall, ask what decision is missing — is it a fact you could look up, a preference you have not settled, or something that can only be known by doing the work? Facts and preferences are yours to resolve, and resolving them closes the scope. Only the third case is real discovery.
Where discovery is real but small, the honest structure is two contracts rather than one: a short paid diagnostic that produces a written finding, then a closed fixed-price job built from that finding. You pay for the unknown once, at a bounded size, and the build gets priced against a list rather than a hope. It also gives you a low-cost read on whether you want to keep working with this person, which a single large engagement does not.
The one thing not worth doing is asking several sellers to quote a fixed price against an open brief and treating the spread as market information. The numbers will differ mostly in how much risk each seller priced in, which tells you about their appetite rather than about your job.
Milestones, and what they actually buy you
A fixed-price job split into milestones is still a fixed price. It does not make an open scope safe; it makes a closed scope legible.
Milestones earn their overhead when the job has real dependency structure — where getting stage two wrong wastes stage three, and stage two is checkable on its own. A theme build where the section structure is agreed before anything is styled. A photo set where one product is finished and approved before the remaining seven are produced. An email flow where the step table is settled before any copy is written.
They are overhead you can skip when the deliverable is single and indivisible. Splitting a logo commission into "concept" and "files" mostly buys you two waits and two review cycles for one artefact, and it invites the seller to treat the concept stage as the negotiation it should not be.
Set each milestone so that it has an acceptance test you could apply without the seller present. "Structure agreed" is not a milestone. "Section list confirmed in writing, in order, with the content each section will hold" is one.
Routing table
| Job | Shape | Why | What closes it |
|---|---|---|---|
| Product image set | Fixed price | Countable artefacts in named formats; delivery checkable against your list. | Product count, shot types, formats, background rules, one lifestyle composite or none. |
| Logo and brand kit | Fixed price | You are buying a file set, not a picture. | Variants, source files, colour spaces, licensing, avatar and monochrome tests. |
| Product page copy | Fixed price | Known page structure, known number of pages. | Page count, section order, claims you supply, claims nobody may invent. |
| Ad cut from supplied footage | Fixed price | Inputs exist; output is a defined number of cuts at defined lengths. | Footage handed over, hook fixed or open, aspect ratios, cut count, caption style. |
| Welcome and abandoned-cart flows | Fixed price | Sequence length and trigger logic are decidable in advance. | Step table: trigger, delay, subject, goal, and who writes each. |
| Theme or landing page build | Fixed price with milestones | Enumerable once sections are named, but stage two wasting stage three is real. | Section list in order, content per section, then styling as a second milestone. |
| Pre-launch QA pass | Fixed price | The deliverable is a defect list against a stated device and flow matrix. | Devices, browsers, flows to walk, and the report format you will accept. |
| Unexplained checkout drop | Hourly, capped | The first task is to find out what the task is. | Nothing yet — buy the diagnostic, then price the fix from its finding. |
| Inherited store assessment | Hourly, capped | You are buying judgement, not an artefact. | A bounded first block plus a written finding you can act on. |
| Ongoing small fixes after launch | Hourly with a review rhythm | Maintenance is availability over time, not a parcel. | An hour cap per period and a standing rule for what needs approval first. |
Brief
ROUTING THIS JOB What I want to exist when this is finished, in one line: CAN I ENUMERATE IT? Deliverables I can name right now - item, format, quantity: 1. 2. 3. Where my list stops, and why it stops there: ( ) a fact I could look up -> look it up, then close the scope ( ) a preference I have not settled -> settle it, then close the scope ( ) only knowable by doing the work -> this part is real discovery SHAPE I AM CHOOSING ( ) Fixed price - my list above is complete and countable ( ) Fixed price with milestones - complete, but stage order matters ( ) Hourly, capped - discovery first, build priced afterwards IF FIXED PRICE Acceptance test I will apply on delivery, without the seller present: Source or editable files required (yes / no): What counts as a revision, and what counts as a new brief: IF MILESTONES Milestone 1 - deliverable, and the test that says it passed: Milestone 2 - deliverable, and the test that says it passed: What stage 2 would waste if stage 1 were wrong: IF HOURLY Hours in the first block, as a cap: The written finding I need at the end of that block: The decision I will make when it arrives (continue / redirect / stop): What needs my approval before more time is booked:
What changes about vetting
The three checkable questions do not change with contract shape — same exact job before, brief read, terms agreed. What changes is which evidence carries weight.
For a fixed-price order, the samples matter most, because you are buying a specific artefact and their portfolio either contains it or does not. For hourly, samples matter less and the reply to your pre-order question matters more, because what you are actually buying is how this person reasons when the answer is not yet known. A seller who asks a clarifying question before quoting is a good sign on a fixed price and close to a requirement on an hourly contract.
One shape-specific check is worth adding to the hourly version: ask how they report time and what a written update contains. You are not auditing their honesty, you are finding out whether updates exist by default or only when you chase them. The answer arrives in one message and it is the single best predictor of whether the review rhythm will hold.
Keeping the transaction on the platform matters more on hourly than on fixed price, not less. A fixed-price order leaves a compact record — brief, delivery, revisions. An hourly engagement's record is the thread itself: the redirects, the approvals, the moments you agreed to change direction. Move that off-platform and the only account of what was authorised is two people's memories.
The mistake worth avoiding
Buyers route by cost anxiety. Hourly feels dangerous because the total is unknown, so an open job gets forced into a fixed price where the unknown does not disappear — it just moves into the seller's number, or into the delivery you did not want.
Fixed price feels safe because the number is knowable, which is true, and irrelevant when the list behind the number is not. A knowable price against an unknowable scope is the most expensive arrangement on this page, because you pay it twice: once for the delivery that matched your vague brief, and again for the job you actually needed.