Reading the commercial terms behind the ways to buy testing work

A testing proposal can describe the same service under very different commercial terms. Two vendors offering the same automation framework can structure the engagement as a fixed-price project, a recurring subscription, or a standing team billed by the month, and the difference changes who manages the work day to day, who invoices whom, and what happens when the volume of work changes partway through the term. The catalog records nine such arrangements. This page works through what sits behind each one and ends with a way to match a model to a specific need.

Five questions that separate one label from another

Before comparing two vendors that list the same engagement model, the model itself is worth unpacking along five lines: who directs the work on a daily basis, who is the legal employer of the people doing the testing, how the bill gets calculated, what length of commitment the buyer is signing up for, and what notice a buyer has to give before changing the size of the work. A proposal that names the wrong label, or leaves one of these five points unstated, can hide a very different set of obligations than the label on its own suggests.

Two models where the vendor stays the legal employer

Two of the nine labels keep testing staff on a vendor's own payroll while handing daily direction to the buyer, in whole or in part. Under the label commonly called staff augmentation, the buyer assigns tasks and sets priorities each day, while the vendor remains the employer, runs payroll, and replaces anyone who leaves the assignment. Billing tracks time worked, invoiced against an hourly or daily rate. Under the label commonly called dedicated team, the vendor typically keeps a layer of day-to-day management in place, with a lead who reports on progress to the buyer, while the buyer still sets priorities and direction. Billing here is usually a recurring fee sized to the roster instead of hours logged by each person individually. Both labels commonly run on a stated contract term, often quarterly or annual, with a notice period for reducing headcount written into the contract. A question worth asking before signing either one: who has the authority to reassign or replace a specific tester mid-contract, since that authority determines how much daily control the buyer actually holds.

Managed testing service: an outcome delivered on a recurring fee

Under a managed testing service, the vendor plans the test cycles, decides which tests run each cycle, and reports results against a service level instead of against a headcount. The buyer states what needs coverage; the vendor's own management decides how many people work on it and how that work gets divided week to week. Billing is typically a recurring fee tied to a defined scope of coverage, adjusted when the scope changes rather than when hours logged go up or down. Because the vendor owns the operational plan, a buyer starting this model is committing to a full cycle, often a quarter, before the vendor's staffing and reporting rhythm can be judged. Changing scope inside a live cycle usually needs notice measured in weeks, since staffing and scheduling have already been built around the prior scope. A question worth asking before signing: what happens to the fee and the reporting cadence if coverage needs to pause for one release cycle without ending the contract.

Project work: one price for one defined scope

Project work bounds an engagement to a specific piece of testing with a start date, an end date, and usually a fixed price or a capped estimate. The vendor manages its own staff for the duration; the buyer's involvement is typically limited to defining acceptance criteria and reviewing the delivered result against them. There is no standing commitment beyond the stated scope: once the deliverable is accepted, no ongoing obligation remains for either party unless a new project gets agreed separately. Because the price is fixed to a scope, the effective notice period for a change is whatever the change order process requires, and a mid-project change to what is being tested commonly resets both the price and the timeline. A question worth asking before signing: what specifically triggers a change order, and who decides whether a requested change falls inside or outside the original scope.

Crowdtesting: the vendor as a directory of independent testers

Crowdtesting routes work to individual testers who are not employees of the vendor and are not directed day to day the way an assigned team would be. The vendor operates the platform, sets the test brief, and selects testers from its own pool, but each tester typically works on a single test cycle as an independent contractor instead of being staffed onto a standing assignment. Billing commonly ties to results delivered, such as a confirmed defect report or a completed test session, instead of hours logged by a named person. A single cycle usually carries no long-term commitment, and a buyer can commonly start or stop one with limited advance notice, since no dedicated headcount has to be reassigned elsewhere once the cycle ends. A question worth asking before signing: how a submitted defect gets verified before it counts toward the fee, since a volume-based payment model rewards submissions unless a verification step sits between submission and payout. The crowdtesting ranking lists vendors that offer this model.

On-demand testing: capacity bought only when it gets used

On-demand testing sells testing capacity in units that can be requested as needed instead of reserved in advance on a fixed schedule. The vendor keeps testers or automated capacity on standby and allocates it to a request as one arrives, so daily direction of any single request sits with the vendor's own dispatcher rather than a manager assigned permanently to the buyer's account. Billing is usually per unit of work requested, such as a test cycle or a defined test pass, invoiced after the fact. Because no ongoing headcount is reserved, the arrangement itself typically carries no minimum term, though an individual request, once accepted and started, usually cannot be cancelled without still owing for the unit requested. A question worth asking before signing: what turnaround commitment applies once a request is submitted, since paying only for work actually done matters less if no response time is stated alongside it. The on-demand testing ranking lists vendors offering this model.

Continuous testing program: a standing arrangement tied to release cadence

A continuous testing program ties testing work directly to a buyer's own release cadence instead of to a fixed scope or a single request. The vendor typically embeds a process, and sometimes a fixed team, into the buyer's pipeline, running tests on every build or every release. Because the arrangement follows an ongoing cadence instead of a bounded deliverable, billing is usually a recurring fee sized to expected pipeline volume, with adjustments negotiated when release frequency changes materially. This model asks for the longest standing commitment of the nine, commonly a year or more, since the vendor builds its process and tooling specifically around the buyer's pipeline. Reducing scope inside a live term typically needs the longest notice period among the nine models, since removing a step from an integrated pipeline changes tooling for both parties. A question worth asking before signing: what happens to the fee if the buyer's own release frequency drops for a quarter, since a program billed on expected volume is exposed to a swing that neither party fully controls. The continuous testing program ranking lists vendors that offer this model.

Two labels that describe how work is organised, not what is delivered

The two remaining labels sit at a different level from the rest. A testing center of excellence is an arrangement where the vendor builds a shared testing function serving several of the buyer's own teams at once, setting common tooling, common reporting and common standards across them. Its output is a way of working, so the questions that decide whether it succeeded are about adoption inside the buyer's organisation instead of about a delivered test run. Billing usually mixes a standing fee for the function with per-team costs, and the commitment is long by nature, since a shared practice cannot be assembled and dismantled inside a quarter.

Time and materials names a billing basis instead of a delivery arrangement, and that is why it can appear alongside almost any other label on the same proposal. Hours actually worked are invoiced at agreed rates, with no fixed scope commitment attached. It gives a buyer the most freedom to change direction mid-engagement and gives the least certainty about the final figure, so the clause worth reading closely is the one governing estimates: whether the vendor commits to warn the buyer at a stated threshold before the running total passes the budget, and in what form that warning arrives.

Neither of these two labels currently has a ranking page of its own, because too few profiles in the catalog record them for a ranking to be published.

Matching a model to the task at hand

None of the nine labels is a default choice; each answers a different combination of the five questions above. A task with a hard deadline and a bounded scope fits project work more closely than a standing program, since a program is priced around an ongoing cadence that a single deadline does not need. A task that needs a large volume of device or locale coverage for a short window fits crowdtesting or on-demand testing more closely than a dedicated team, since neither carries a standing headcount left idle once the coverage window closes. A task that needs long-term ownership of a regression suite, with the same people carrying product history from one release to the next, fits a dedicated team or a managed testing service more closely than a project, since a project's obligation ends once its scope is accepted. Once a candidate model is picked on that basis, the same five questions apply again to the specific vendor's contract draft: the label on a proposal states an intent, and the clause on management authority, billing basis, term length, and notice period is what a buyer actually agrees to.