Assessment Platform RFP Checklist for Training Providers

An assessment platform RFP works when it turns your needs into requirements a vendor can pass or fail, and when it tells vendors how you will score the answers. For a training provider, that means a short list of gates, a weighted list of scored requirements, and a rule that every claim comes with evidence.

This checklist gives you the requirements to copy, a response-code scheme, illustrative weights and a worked scoring example with the arithmetic shown. Every weight and number in the examples is an assumption for you to change, not an industry benchmark.

It is written for training providers and L&D managers who write or score the RFP. If you are still deciding whether to buy or build, start with the white label versus building your own exam engine framework. If you are still learning what these tools do, read what an online assessment platform is first.

Key takeaways

• Split requirements into a few pass/fail gates and a longer scored list, and publish the weights in the RFP.
• Make vendors answer each row with a response code (standard, configurable, paid custom, third party, roadmap, not available) and cap the score each code can earn.
• Write standards as a name, a version and a proof: for example QTI 3.0 export, WCAG 2.2 Level AA with a dated conformance report.
• Score capability before prices, then add cost points, because a low bid can hide a large capability gap.
• Test finalists with your own sample items and an exit drill before you sign.

What does an assessment platform RFP need to contain?

An assessment platform RFP needs seven parts: your background and volumes, pass/fail requirements, scored requirements, a required response format, the evaluation method with weights, commercial and exit terms, and a timeline. Vendors answer well only when they know your exam volumes, your learners and how you will judge them.

Write each part so it can be checked:

• Background and volumes: learners per year, the largest cohort sitting at once, exam stakes, and delivery settings such as in-room, remote or phone.
• Gates and scored requirements: the two lists explained below.
• Response format: one spreadsheet, one code per requirement, and a link or file for each piece of evidence.
• Evaluation method: the weights and the scoring scale, stated up front.
• Commercial terms: pricing on your volumes for three years, overage rates, renewal limits and exit terms.
• Timeline: question deadline, response deadline, demo window and decision date.

Public buyers have a rule for the weights point. As of October 2026, the U.S. Federal Acquisition Regulation (FAR 15.304) says the evaluation factors and significant subfactors, and their relative importance, “shall be stated clearly in the solicitation.” A training provider is not bound by that rule. Following it is still fair to vendors, and it gives you a record if a losing bidder asks why.

How do you build the shortlist before the RFP goes out?

Build the shortlist from each vendor’s own published pages, and record “not published” as a question for the RFP, not as a failure. Public pages show what a vendor chooses to say. They are a first filter, not proof, and the RFP exists to turn silence into written answers.

Take SimpliAssess as an example. As of October 2026, its features page lists item banking, test management, examinee management, scheduling, analytics and reporting, bulk import and an API-first design. It does not mention SSO, standards support or proctoring, and its FAQ says users can export all test scores into a spreadsheet without naming item-bank export formats.

None of that is a verdict. Each gap is a row in the requirements table below, and every vendor on your list, this one included, answers those rows in writing. Apply the same “published, not published” test to every site you read.

Which requirements are pass/fail gates, and which are scored?

A gate is a requirement where a “no” means you cannot run your business on the platform, so the vendor is out whatever its other scores. Everything else is scored. Keep gates to five to eight items, or the scored list stops mattering.

Typical gates for a training provider are:

• full item-bank export in an open format, shown with a sample file;
• a dated accessibility conformance report for the learner-facing interface;
• written security and data protection documentation;
• SSO, if your clients require it;
• a documented API for items and results;
• written data export and deletion terms on exit.

Next, make vendors answer every row with a code, so a “yes” from one vendor means the same as a “yes” from another. The code also caps the score, because a roadmap promise is worth less than a shipped feature.

Score each row from 0 to 4: 0 for no answer or no, 1 for an unproven claim or roadmap, 2 for custom work or a third-party add-on, 3 for a documented standard feature, and 4 for a standard feature shown working in your demo or pilot, with independent evidence where it exists.

Which requirements belong in an assessment platform RFP?

Ten areas cover most training providers: content portability, accessibility, security and privacy, delivery and exam integrity, authoring and item banking, reporting, integration, branding, support, and commercial and exit terms. Write each as a testable sentence, and name the evidence you will accept.

For the delivery row, proctoring for training providers explains which integrity controls fit which stakes. For the authoring row, the item bank fields and versioning rules give you tags and statuses to require.

How do you write standards requirements so vendors cannot dodge them?

Name the standard, the version and the proof. “Supports QTI” is a claim. “Exports QTI 3.0 and is listed in the 1EdTech directory with Export certification” can be checked. Do the same for accessibility, LMS launches and reporting formats.

• QTI: 1EdTech describes QTI as a way of packaging and moving tests and questions from one application to another, and names QTI 3.0 as the current specification. It certifies applications for authoring, import, export, import and export, or delivery, and its TrustEd Apps Directory is the official listing of certified products.
• LTI: 1EdTech describes LTI as a technical standard for connecting learning tools to a learning environment, with LTI 1.3 as the current version and LTI Advantage adding grades, rosters and deep linking.
• WCAG: WCAG 2.2 is a W3C Recommendation with three conformance levels, A, AA and AAA. It adds nine success criteria to WCAG 2.1 and does not deprecate 2.0 or 2.1, so check which version your client contracts name.
• Accessibility reports: the VPAT from the Information Technology Industry Council has a WCAG edition built on WCAG 2.2. Completed for one product, it becomes an Accessibility Conformance Report that rates each criterion as supports, partially supports, does not support or not applicable. There is no pass/fail scale, so read the partial rows.

Pro tip: Write Requirements As Tests. Replace adjectives with an action you can watch. “Strong reporting” cannot be scored; “export item-level results for last quarter’s largest cohort and open the file in our spreadsheet tool” can. Each row in your RFP should end in something a scorer sees happen.

How should you weight and score the responses?

Give each area a weight, make the weights sum to 100, and set them before any proposal arrives. Score each requirement from 0 to 4, average the scores within an area, then multiply the average by the weight and divide by 4. Changing weights after you see results invites bias.

The weights below are illustrative for a training provider that sells exams under its own brand. Treat them as a starting point and raise or lower them using the right-hand column.

What does a worked scoring example look like?

Three invented vendors show how the method behaves. Capability points are weight times average score divided by 4, out of 80. Cost points are 20 times the lowest bid divided by the vendor’s bid. All scores and bids are made up for this example.

Two things stand out. Vendor B has the weakest capability score, 8.25 points behind C, yet its low bid brings the totals within 0.25 points. The gate removes it anyway, because a platform you cannot leave is not cheap.

Vendors A and C finish 0.5 points apart, which is noise at this level of precision. Treat gaps this small as ties and decide with the pilot and reference calls. That threshold is a judgment call for this example, not a rule.

Pro tip: Score Capability Before Prices. Have evaluators finish the capability sheet without seeing bids, then add cost points afterward. That way a low bid cannot soften how you read a weak answer, and a high bid cannot harden it.

Which vendor questions expose weak answers?

Good questions ask for an artifact, a number or a demonstration, so a vague answer shows. These eight work for any vendor, and each lists what a weak answer sounds like.

1. “Export our complete item bank and show us the file.” Weak: “Yes, via CSV,” or “we can do that after you sign.”
2. “Which QTI version and certification type do you hold?” Weak: “We support QTI,” with no version or directory listing.
3. “Send your accessibility conformance report for the learner interface, with date and WCAG version.” Weak: a one-line statement, or a report for a different product.
4. “What happens to an attempt when the connection drops mid-exam?” Weak: “That is rare,” with no documented behavior.
5. “Which of your answers are paid custom work or third-party? Price them.” Weak: “Yes” on every row and no prices.
6. “Who inside your company can see our data, and how is access logged?” Weak: general assurances with no process or audit report.
7. “How do we leave?” Weak: export only on request, fee unstated, no written deletion confirmation.
8. “Show an item-level results export from a real exam.” Weak: dashboards and screenshots, no raw file.

How do you test finalists before you sign?

Test finalists with your own material, because demos use clean data. A short pilot of five tasks moves answers from “documented” (3) to “shown working” (4), or exposes the gap. Run the same tasks for every finalist.

Step 1: Import your sample. Send each finalist the same 25 items, including images, math and your messiest formatting, and compare what comes back.

Step 2: Build and brand an exam. Ask someone on your team, not the vendor, to author two item types, set a time limit and apply your branding.

Step 3: Take it two ways. Complete the exam on a phone and again with only a keyboard, and note anything that blocks you.

Step 4: Pull the results. Export item-level results and open them in your spreadsheet tool, then launch the exam from your LMS if you require LTI.

Step 5: Run the exit drill. Export everything you created and ask to see the written deletion process.

What should you do next?

Copy the requirements table into a spreadsheet, delete rows you do not need and add your volumes. Choose your gates, fix your weights, and share the score sheet with every scorer before any demo. Agree that nobody changes weights after proposals arrive.

Then send the RFP to your shortlist, hold one question window so every vendor hears the same answers, and run the finalist tests. Keep the score sheet and write one paragraph on why the winner won, so the decision survives questions later.

If your RFP also covers a learning platform, SimpliTrain, a separate training platform from the company behind SimpliAssess, publishes a free LMS RFP template and vendor questionnaire. Its page says both downloads sit behind a short form and are built around LMS selection, so borrow the commercial and security sections and keep the assessment rows from this checklist.

Frequently asked questions

What is an assessment platform RFP?

An assessment platform RFP is a request for proposal that asks vendors to answer your requirements in writing, with pricing, so you can score them the same way. It usually covers item banking, delivery, security, accessibility, reporting, integrations and exit terms. An RFQ, by contrast, focuses mainly on price.

Do small training providers need a formal RFP?

Not always. If you are comparing two or three tools for one course, a trial and a written quote may be enough. A written RFP pays off when exam results carry stakes, several people must agree, you will hold the item bank for years, or a client or accreditor may ask how you chose.

How many vendors should receive an assessment platform RFP?

Send it to as many vendors as your team can score properly. A shortlist of three to five fits most small teams, because scorers must read every response and watch every demo. This is practical judgment, not a rule. More vendors mean more reading and usually weaker scoring.

What standards should an assessment platform RFP require?

Common choices are QTI 3.0 for item portability, LTI 1.3 for LMS launches, and WCAG 2.2 Level AA with a dated conformance report for accessibility. Add xAPI or SCORM only if your LMS needs them. Ask for proof, such as a 1EdTech directory listing, rather than accepting a yes.

What is a VPAT, and should vendors supply one?

A VPAT is a free template from the Information Technology Industry Council. A vendor completes it for one product, which turns it into an Accessibility Conformance Report rating each criterion as supports, partially supports, does not support or not applicable. Request the WCAG edition, with date and product scope, and read the partial rows.

How do I avoid lock-in with an assessment platform?

Require a full item-bank export in QTI, tested on your own sample before you sign. Also ask for a separate export of tags and version history, since you should not assume a standard carries every field. Put export and deletion terms, with time limits, and a renewal price cap in the contract.