Skip to content
NEW: Martech Stack Builder on your phone →

An RFI builder that runs the whole RFI, not just the document

Writing the RFI isn’t the slow part. Chasing vendors, collecting answers from five inboxes and getting three colleagues to score the same thing the same way… that’s where the weeks go. So Stack Builder runs the RFI in one room, from the first draft to the decision, and keeps the vendors’ answers, the scores and the reasoning together.

Open the Canvas See pricing

No credit card required

From one sentence to a recorded decision

Describe the need, confirm one page, send it to vendors, see who has answered, score every answer side by side and record why you chose. The decision then starts a change request, already filled in.

An RFI run in the RFI room. The need is typed in one sentence; the app fills in the capabilities and a shortlist, and generates the RFI with editable sections. It is sent to vendor contacts with one deadline, the room shows who opened it and who has answered, a vendor that asked for more time is given its own later deadline, one question is scored for every vendor side by side from their answers, and the scores are agreed and the decision recorded with its reason on one page. The decision becomes a pre-filled change request, which is reviewed, applied and tracked, and the Inbox lists what is waiting on you.

The ROI is hard to ignore

A typical agency runs 4+ vendor evaluations per year. Here is what the manual process actually costs.

Manual process

4 RFIs/year × 25 hrs × $200/hr

$20,000/year

With Stack Builder

Consultant plan, billed yearly, excl. VAT — unlimited RFIs

$990/year

100+

Hours saved per year

$19K+

Cost saved per year

22x

ROI multiplier

The manual RFI problem

Every vendor evaluation cycle drains 17-32 hours of skilled consulting time.

Scoping requirements from scratch

2-4 hrs

Writing the RFI document

3-6 hrs

Distributing & collecting responses

4-8 hrs

Building a scoring matrix

6-10 hrs

Creating the recommendation deck

2-4 hrs

Total per evaluation cycle

17-32 hours

Five stages, one room

Every RFI moves through Draft, Send, Collect, Score and Decide. The room shows which stage it’s in and the one thing it’s waiting for next, whether that’s sending, fixing the weights or agreeing a score.

  1. Draft

    Describe what you're buying in a sentence. The engine matches it to capabilities, shortlists two to five vendors from the catalog and fills every topic it can from the stack you've already mapped. You confirm one page, and the AI writes the RFI with a weighted scoring rubric.

  2. Send

    Pick a contact at each vendor, set one deadline and send.

  3. Collect

    Watch the answers come in, vendor by vendor and question by question.

  4. Score

    Score every vendor against the rubric, one question at a time, side by side.

  5. Decide

    Pick a vendor, or none of them, and write down why.

Every RFI in one list

The RFI library lists every RFI in the workspace with its stack, its stage, the vendors, the due date, the owner and the next action. Filter it to what's open, decided or closed, or to the ones that need you.

The RFI library with four RFIs on one stack, one at each point: closed with OneTrust chosen, deciding, collecting answers and ready to send, each with its vendors, due date, owner and next action

Draft: from one sentence to a document you can edit

Start an RFI by saying what you're trying to buy, in a sentence, the way you'd say it to a colleague. The engine reads it against the catalog, turns it into capabilities and shortlists the vendors that have them. Your stack, your company and who is asking are already known, so it doesn't ask for them. A phrase the catalog can't place isn't dropped: it becomes one question to the vendors.

Then you confirm one page: the capabilities, the vendors (pick two to five) and anything optional you want to add, such as category questions from the bank for that category or your own. Each topic is marked as filled by the engine, filled by you, recommended, optional or needed before it can be generated, so you know what's left. Prefer to walk every topic yourself? Build it manually instead; nothing you typed is lost.

After generation, every section is editable. Reorder sections, leave some out, and set each question to required or optional. Preview it the way the vendor will see it before you send. Export it as a Word file, print it to PDF or copy it as Markdown.

A new RFI on the Northwind stack: the need typed in two sentences under “What are you trying to buy?”, and below it what the engine already knows and will not ask: the stack, the company and who is asking
The one page to confirm: the need in the person’s own words, three capabilities from the catalog, the unmatched phrase about the OneTrust consent signal turned into a vendor question, and the vendor shortlist with mParticle, RudderStack and Twilio Segment ticked, each showing how many of the capabilities it covers
Confirm one page, then generate the RFI

Send and Collect: no more chasing by hand

One send dialog covers every vendor: the contacts, one deadline with its time zone, and a preview of the email they'll get. Reminders go out on their own: seven days and one day before the deadline, and to anyone who hasn't opened the RFI after a few days. You can remind someone yourself too. Either way, nobody gets more than one every 48 hours, and the reminders stop as soon as a vendor submits.

Did the email even arrive? Collect shows where each person stands: sent, delivered, opened, answering, submitted or bounced. A bounce shows up as a bounce, not as silence.

The Send to vendors dialog: one contact ticked at each of Braze, Klaviyo and Mailchimp, the answer deadline and its time zone, where replies go, the automatic reminders, and a preview of the invitation email
An RFI at the Collect stage: each vendor contact with their own timeline (invite sent, delivered, opened the email, opened the RFI, downloaded the workbook), a Remind button, and a log of what happened, newest first
Collect: where each person stands

What the vendor sees

Vendors don't need an account. They get a personal link to a page with your questions, and they answer question by question, with evidence and a link where you asked for it. Drafts save as they type, so they can finish on another device with the same link. The page tells them what's still missing before they submit.

Some vendors would rather answer in a spreadsheet. (Usually the ones with a bid team.) They can download the questions as an Excel workbook, fill it in offline and upload it. Before anything is saved, they see what was read from the file and what wasn't. Uploading doesn't submit; they still review and send.

The answer page a vendor sees, for the fictional Meridian CDP: 23 of 41 items answered, the sections with what is left in each, the option to download the questions as an Excel workbook or upload a filled one, and a requirement answered as Available with its evidence
The check after a vendor uploads a filled workbook: 4 new answers, 3 that would replace an answer given online, 16 the same, 3 that cannot be used and 2 rows not matched, with the online answer and the file’s answer side by side and nothing saved until the vendor ticks it
An Excel upload, checked before anything is saved

Score: side by side, then agree

Scoring opens as soon as the first vendor answers. The split view shows one question at a time with every vendor's answer next to it and the weighted rubric beside them. Scores save as you go.

Colleagues score with their own account. People outside the workspace can score through a link, without an account. On a team RFI, the consensus grid starts each score at the scorers' average. And it flags any question where they're two points or more apart, so you talk about those and agree the rest. Scores lock when the RFI closes.

The Score stage split view: one question at a time, with its section and weight, the fictional Meridian CDP’s answer next to two vendors that have not answered yet, and a 1 to 5 score for each vendor

Decide, then hand it over

The decision sheet records the vendor you picked, or that you picked none, and the reason. You can tell the vendors the outcome from the same place. Closing revokes the vendor links and stops the reminders, and the decision can be reopened within 14 days.

Then the part that usually gets lost: on Consultant and up, one click starts a change request from the decision, with the chosen vendor, the reason and the scorers suggested as approvers.

Recording the decision: RudderStack chosen at an agreed 86.3% over mParticle at 70.0% and Snowplow at 60.0%, the reason written in, the option to tell the vendors the outcome, and what closing does (lock the scores, revoke the links, stop the reminders)

Who this is for

Built for anyone who needs to evaluate vendors professionally.

Agencies

Deliver professional vendor evaluations to clients in a fraction of the time. Add your own questions next to the category bank, and keep each client’s RFIs in that client’s workspace.

Independent Consultants

Look like a large firm. Generate polished RFI documents and scored recommendations that justify your advisory fees.

Procurement Teams

Standardize your vendor evaluation process. Weighted scoring and team consensus tools ensure objective, defensible decisions.

The method

Run an RFI by hand

An RFI is only as good as the comparability of what comes back. Free-text answers cannot be compared; that is the whole problem this procedure solves. Three to four weeks end to end.

  1. Write requirements as questions with a fixed answer shape

    Each requirement gets a question and a stated answer format — yes/no, a number, or one of a list. "Describe your approach to identity resolution" returns six pages you cannot score.

  2. Publish the scoring rubric with the questions

    Tell vendors what a 5 looks like versus a 3. You get better answers, and you commit to the standard before you know who met it.

  3. Give every vendor the same deadline and the same contact

    Staggered deadlines advantage whoever answered last. One date, one inbox, no side conversations.

  4. Score independently, then meet

    Each scorer scores alone before anyone compares. Group scoring converges on the loudest person in the room within about ten minutes.

  5. Discuss only where scores diverge by two or more

    Agreement needs no meeting. Divergence is either a misread question or a genuine difference in what people value — both worth the time.

Worked example Illustrative

CEP selection, four vendors, 34 scored questions, three scorers.

Questions issued 34
Vendors responding by deadline 3 / 4
Questions with ≥2 point divergence 7
Divergence from misread question 4 of 7
Score change after discussion 2 vendors

Reading it: Four of seven disagreements were a badly worded question, not a difference of opinion. That ratio is the argument for step 1 — most scoring disputes are authoring defects.

The shortcut: The RFI builder gives every criterion a fixed answer type (a 1–5 scale, yes / no / partial, or structured fields), and colleagues score side by side in the same room (outside scorers through a link).

The argument behind this

The CDP RFP dating game

On what the RFP process rewards, and what it hides.

Frequently Asked Questions

How much does the RFI builder cost?
Pro includes your first RFI. Consultant ($99 a month), Agency ($499 a month), Team ($35 a seat a month) and Enterprise include unlimited RFIs. The free plan doesn't include RFIs.
Do vendors need an account?
No. Each contact gets a personal link. They can answer on the page or in an Excel workbook, and their drafts are saved on our side until they submit.
Can my team score vendors together?
Yes. On a team RFI, each colleague scores on their own and the consensus grid shows where you disagree. On any plan with RFIs you can also invite outside scorers by link, without an account.
What export formats are supported?
The RFI document exports as a Word file, prints to PDF from your browser or copies as Markdown. Scoring results export as Markdown, CSV, a PNG of the comparison table, or a full report as Markdown, HTML or PDF.
How does the AI write the RFI?
You describe what you're buying in a sentence. The engine matches it to capabilities, shortlists vendors from the catalog and fills every topic it can from the app. You confirm one page, and the AI writes the document from your stack and the category questions for that kind of tool.
How many vendors can one RFI include?
Two to five. More than five and nobody scores the last two properly anyway.

Related Solutions

Vendor Evaluation

Compare vendors with weighted scoring and team consensus tools.

Learn more

Contract Management

Upload vendor contracts, extract the key terms, and ask questions with clause citations.

Learn more

Stack Audit

Overlaps, an A–F grade and recoverable spend, computed from your stack.

Learn more

Stop chasing vendors by email

Draft the RFI from your stack, let the reminders do the chasing, and score every vendor side by side. One room, from the first draft to the decision.

Open the Canvas See pricing

No credit card required. Free plan available.