Rightmark vs AWS Data Exchange: a 2026 head-to-head comparison

Compare the work each route covers, then examine the proposed asset, agreement and receiving workflow.

Before choosing: Confirm that offers, fees, approvals, permitted use and delivery match the current proposed agreement.

Start with the risk of buying data that does not do the job, then examine the money, permissions and work needed to use it. The seven questions below keep the owner's licensing decision beside the buyer's acquisition decision.

1. Will the data meet the buyer's actual requirement?

A sample can look convincing while leaving the hardest part of the task untested. Buyers need to know what is missing as well as what is present; owners need a scope they can deliver without promising material they do not control.

Make the sample answer a question that could change the purchase. If your task depends on a sequence of events, a handful of isolated records will not test whether that sequence survives preparation. Request a permitted sample that preserves the relevant relationships, accompanied by an explanation of what was removed. The owner should be able to trace that sample back to the material covered by the proposed license.

Keep a short acceptance record: required fields, required history, exclusions, the test to run and the person who decides. Record unknowns openly. The result may be to narrow the collection or change the task; it should not become an unexplained pass because the file opened successfully. The buyer request brief is a starting point for writing those requirements.

Before committing, put the acceptance test and mismatch remedy in the offer, including who decides whether the delivered scope conforms.

Bottom line: Evaluate Rightmark when finding and preparing the private collection is the work still to do. A suitable provider product makes AWS worth testing directly. Ask for the same evidence in either case: a sample, its exact scope and an acceptance test.

What to ask both: "Can you show a representative sample tied to the exact licensed scope, the known exclusions, and the test and remedy we will use if delivery differs?" A volume or grade without a definition of what will be accepted is a reason to keep asking.

2. What will the owner receive, and what will the buyer spend?

The owner cares about the amount retained after costs. The buyer cares about the complete budget needed to obtain and use the data. Comparing one route's owner payout with another route's invoice makes the cheaper-looking number meaningless.

Our published standard-license rule is: "We take nothing from your payout on standard licenses; the buyer pays our fee on top." Premium and bespoke deals have a different arrangement: the homepage states an owner share of 75% to 85%, agreed in writing. That distinction belongs in the quote. Keeping the agreed payout does not mean keeping the buyer's entire spend.

AWS's pricing separates delivery costs. Uploaded files incur provider storage charges measured in byte-hours, at the region's rate. Direct S3 avoids that upload; the provider still pays S3 storage. Enable Requester Pays and requesters pay for requests and downloads; disable it and the provider can incur those charges.

Direct S3 access lets the buyer use the provider's existing objects without keeping another copy. For a buyer who needs the current version, price that route before paying to store and maintain a duplicate. If you need a separate retained copy, put that requirement in the scope.

Use the worksheet's transfer, storage and processing row for each charged service. Record its rate, billable quantity and period, then name the payer. Byte-hours means bytes multiplied by hours. Put an owner-paid item against owner net and a buyer-paid item against the buyer budget; an unassigned charge leaves both totals unfinished.

For a fair comparison, write one specification and ask both sides to price it. Keep the same permitted use, date range, term, acceptance test and delivery requirement. A broader license or extra preparation can justify a higher figure, but only if you need it. A low price for a scope you cannot use does not solve the buying problem.

Use this blank worksheet with actual quoted inputs. Keep one-time and recurring amounts separate, and note the currency and tax treatment. Do not put private data or a counterparty's confidential terms into a public form.

Buyer budget is the license or product charge plus separately charged buyer fees, preparation, transfer, storage, processing and applicable taxes. Owner net is the agreed receipts less applicable owner-paid fees and work. Count each item once: an included transfer charge is not added again. For usage costs, multiply the quoted rate by measured usage or an explicitly labeled usage assumption; keep the assumption visible.

The Rightmark Price Index can help you identify the unit and context behind a reference figure. It is not a competing offer for your asset. The actual scope and quote decide the economics.

Bottom line: Rightmark's standard fee rule lets the owner start from the agreed payout, then identify any retained work. For AWS, direct S3 access is worth pricing when a second stored copy adds no value to the buyer's task. Compare complete budgets before choosing.

What to ask both: "For this scope and term, can you itemize the owner's receipts, every fee and delivery cost, the buyer's total expected spend, and what changes after a rejection or renewal?" A rate without its unit or an amount without its payer is unfinished pricing.

3. What rights and handling obligations follow the data?

The ability to inspect a sample and the right to train a model are different permissions. Before committing, the buyer needs a documented grant for its intended use, and the owner needs to understand which copies, recipients and derived material that grant covers.

Rightmark's ownership-pack model puts the evidence into a form a buyer can review. On the code route, the published sample folder covers contributor and IP documents, never-public evidence, open-source disclosures, scans and a data card. That is useful organization. It is not a legal certification, and the actual folder must support the actual collection.

Ask the reviewer to name what each document establishes. A contributor assignment may address one part of the rights chain while leaving client material or third-party components unresolved. Keep exclusions attached to the scope and sample rather than burying them in email. If a requested use is missing from the agreement, resolve it before using the data for that purpose.

Our public license summary is a draft for counsel review. It explains that some code routes can allow a licensing partner to sublicense, take an own-use license or build derived datasets. Ask for the route's full agreement; do not assume every license has the same resale restrictions or protections.

Draw a custody map with separate entries for the source, evaluation sample, licensed copy and any outputs the agreement covers. Against each, record who holds it, permitted users, location, retention and deletion evidence. Ending a connection answers an access question; it does not, by itself, account for a file already exported. Rightmark's Trust page provides a worksheet for that discussion.

Bottom line: Favor the route that can produce the rights and handling evidence for your exact use. Rightmark's organized packet helps the review; AWS's technical controls help the operating environment. Neither should substitute for the missing document or unresolved permission.

What to ask both: "Which documents establish authority for this exact use, where will samples and licensed copies reside, and who must provide deletion evidence when the agreement ends?" A single answer for all AI uses or all copies needs a more precise scope.

4. Which decisions can the owner and buyer approve?

Control is easier to assess when each decision has a name. An owner should know whether signing approves one license or a route for future licenses, and a buyer should know who can accept its requested use and commercial terms.

The code qualification is consequential. Rightmark's Trust page says the NDA and license are one agreement on code routes. After signing, the owner connects to the licensing partner chosen for the deal; labs can then license the code nonexclusively without a further owner approval. Existing licenses survive withdrawal. An owner who expects to approve every later license must settle that expectation before signing.

AWS should not be treated as a public listing with no private negotiation. Its private-product and private-offer mechanisms give providers a way to control availability and tailor an arrangement. Ask which decisions your provider will actually make, and have the offer reflect them.

Put the approval sequence on one page: material included, permitted sample audience, use, commercial terms, first delivery and later changes. Beside each decision, write the approving role and whether approval is one-time or recurring. This reveals disagreements that a general promise of owner control can conceal. A promise to stop future licensing should be written separately from any obligation concerning copies already licensed.

Before signing, ask what information you will receive about the final licensee and which changes to the material, recipient or use require a new scope decision.

Bottom line: AWS is attractive when you want to manage subscriber access and offers yourself. Rightmark can help document the licensing scope, but code-route signing can authorize later licenses. If you expect to approve every future code license, resolve that requirement before choosing the route.

What to ask both: "Who approves the scope, subscriber and terms, which approval covers later licenses or changes, and what can we withdraw after access or a license has been granted?" Keep future decisions and existing grants separate in the answer.

5. Can the route deliver data in the form the buyer needs?

A relevant dataset can still require substantial work before it reaches the people who need it. The buyer needs a usable transfer path and update plan; the owner needs to understand the ongoing publishing work attached to that promise.

AWS has the clearer advantage when a suitable product is being used in an established AWS workflow. Its data-sharing guidance describes file, Redshift-table and API delivery alongside consolidated billing and subscription management. That gives the buyer a concrete operating model to assess.

There is also a documented option for a subscriber without its own AWS account. In AWS's provider-managed setup, the provider creates the account and configures identity and permissions. It is a provider-built arrangement, so ask whether your provider offers it and who will maintain it; it is not the default promise of account-free purchasing.

For a private archive, agree what a delivery actually contains before organizing the transfer. A folder of files, an export with linked records and a continually updated source are different requests. List the schema, date range, exclusions, version and receiving owner. Then decide what happens if a batch arrives incomplete or a later schema change breaks the import. The Rightmark request brief provides fields for format, history, filters and cadence.

Bottom line: AWS wins this dimension when the selected product and your receiving environment fit its delivery paths. Rightmark is worth evaluating when the missing piece is the private collection and its preparation. Confirm supply and the exact transfer scope before planning around it.

What to ask both: "Can you map the exact delivery format, version and update schedule to our receiving system, with the permitted users, locations and person responsible for each step?" A matching category with no concrete transfer plan is not enough.

6. Who does the work before and after the first delivery?

A license can be commercially acceptable and still leave a team with work it did not budget for. Before choosing a route, the owner and buyer should agree who prepares the evidence, who runs delivery and who answers when something changes.

The benefit to an owner is tangible: AWS's Hudi publishing walkthrough explains how providers can use its delivery, entitlement and billing infrastructure while concentrating on their datasets. It also walks through registration, product setup and publication. The owner must be ready to operate that product.

Rightmark's published process covers a different part of the job. Scans run on the owner's side or in a buyer pipeline; we review the resulting reports and scrubbed samples, assemble the route's ownership pack and match the dataset. The owner still needs to supply sound evidence, identify exclusions and approve access. The buyer still needs someone able to evaluate the sample and accept the delivery.

Before signing, assign a named role to every task that matters: scope, authority evidence, preparation, delivery, receipt testing, updates, support and deletion. Mark each as included in the quote, separately charged or retained by your team. Pay particular attention to the handoffs. If one team prepares the sample and another delivers the full collection, decide who checks that both use the same scope and version.

Have the proposal name who handles integration, updates and continuing support, with any charges and response commitments stated explicitly.

Bottom line: If you can maintain the dataset and its product terms, AWS can handle distribution, entitlement and billing. Choose Rightmark when you need reports reviewed, a licensing packet assembled and a buyer matched; your team still supplies the rights evidence and approves the scope. The buyer retains the sample test and delivery acceptance.

What to ask both: "Who is responsible for scoping, rights evidence, preparation, product setup, delivery, acceptance, updates and support, and which of those tasks are included in this quote?" If each party assumes another owns the same task, resolve that before delivery.

7. What proof matters in an AI training data marketplace comparison?

A published workflow tells you how an arrangement is intended to work. A purchase needs evidence about the collection you would actually receive. Keep technical demonstrations, market reporting and comparable completed engagements separate when judging the strength of that evidence.

AWS's Redshift ML walkthrough is a useful technical example. It subscribes to a test product with event-attendance data, connects to a Redshift datashare, and builds a model to predict ticket sales. The tutorial specifies prerequisite resources and warns that some incur charges. That demonstrates an implementation path; it does not establish permission for every AI use, a production result for your team or the quality of another provider's dataset.

For Rightmark, inspect the actual ownership pack and sample rather than treating the public illustration as your evidence. The homepage expressly says its linked news reports concern the wider market, not Rightmark. They do not prove our placement rate, your likely proceeds or a result from a comparable license.

A useful evaluation ends with a decision someone can defend. Record the sample version, permitted test, observed gaps and acceptance decision. Preserve the failed checks as well as the successful ones. An owner should learn what preparation is still needed; a buyer should learn whether a larger commitment would answer the intended task. A technically readable sample can still lead to a decision to stop.

Bottom line: AWS's Redshift example helps a technical team test the subscription-to-model path; it cannot answer whether the offered collection is licensed for your use. With Rightmark, proceed only when the ownership pack supports the proposed rights and the scrubbed sample meets your acceptance test.

What to ask both: "What evidence from this exact asset and a comparable permitted use can you show, and what would we need to verify in a bounded evaluation before committing?" A demonstration about a different dataset is a starting point for questions.

Our work starts with finding a private collection and assembling its licensing evidence. These five parts of the process matter when you need an owner found or an existing archive prepared for review.

• Private sourcing from a written requirement. Our buyer workflow starts with a specification and describes finding owners whose data has never been listed. The written scope covers families, filters, formats, volume and cadence; availability and timing still need confirmation.

• Evidence organized for the route. The ownership pack brings together documents and reports for counsel to inspect. Its contents depend on the route, so ask what establishes authority for the collection and what remains unresolved.

• Evaluation before the full delivery. Buyers receive reports and representative scrubbed samples under NDA for evaluation only. Our process calls for graded batches with metadata and another scan on receipt; acceptance criteria and remedies need to be agreed for the engagement.

• An explicit agreement and access decision. Owners approve the agreement and scope, with transfers through approved connections or exports. Code-route signing can authorize later nonexclusive licenses without another approval; withdrawal applies to code not already licensed.

• A stated standard-license fee allocation. The buyer pays our fee above the agreed owner payout on standard licenses. Premium and bespoke shares are agreed in writing, and the Trust page says fees are shown before signing.

Choose AWS when you are ready to operate a defined data product, or when you have found a suitable provider product and want to use it in an AWS workflow. Delivery and subscription management are real reasons to choose it. A buyer already equipped to receive the data should assess that fit directly.

For an owner, readiness means having people who can describe the collection, maintain its quality, set terms and support the arrangement. For a buyer, it means knowing the required fields, history, use and acceptance criteria. Pay for additional sourcing or preparation only when it closes a gap your teams cannot cover.

An owner can keep a product out of the public catalog and make it available only to selected customers. Provider verification can restrict who subscribes. A buyer can negotiate a private price, term and agreement within AWS. If both sides already know the dataset and counterparties, private negotiation does not require moving the transaction elsewhere.

Evaluate Rightmark when you have a private codebase or company archive and need help defining what can be licensed, assembling the evidence and arranging the next step. Our published process reviews reports and scrubbed samples and prepares the ownership pack. You still need to establish authority and approve the agreement, including any code-route authorization for later nonexclusive licenses.

For a buyer, the stronger fit is a team with a clear specification and the ability to evaluate data, but no identified private collection that meets it. Rightmark offers sourcing, written scope, sample evaluation and graded delivery. Confirm the collection, permitted use, acceptance criteria, timing and complete quote before treating a proposal as usable supply.

The limits matter. Health-data intake remains unavailable. A lot marked in preparation is not available supply, and an appraisal is not an offer. Standard-license economics should remain separate from premium or bespoke terms. If you need a particular collection immediately, an exact dataset and confirmed delivery matter more than either route's general description.

Ask Rightmark to confirm whether it can take your specific collection or request, what preparation is required, who performs it and what capacity or timing limits apply.

A direct license is worth comparing when the owner and buyer already know each other, the collection and use are defined, and both teams can manage the agreement and transfer. Assign the authority review, sample permission, pricing, acceptance, delivery, invoicing, support and end-of-term work before deciding that the remaining task is small.

Include that internal work in the comparison. A direct arrangement is not automatically cheaper, simpler or safer. If the obstacle is unresolved authority or permission, changing the route does not remove it: resolve the issue or narrow the collection first.

Relevant reading: Rightmark vs Defined.ai.

Start with the work you need done. AWS is a strong candidate for a defined provider product and a compatible receiving workflow. Evaluate Rightmark when finding, scoping and preparing a private collection is the missing work. The two models can serve complementary needs; this comparison does not establish an integration or partnership.

Take the same specification and the questions above into both conversations. Record the scope, permissions, responsible people, acceptance test and complete costs. The owner's receipts and buyer's spend should be clear before either side commits.

Describe the private data your team needs.

Send Rightmark your specification, intended use, volume, budget and timing. Rightmark replies with what it can source, volumes and timing, then puts the proposed scope in writing.

Request data