Technical SEO

Title, meta description, canonical, OG, Twitter, BreadcrumbList, FAQPage, WebApplication schema.

Feature Request Triage Assistant (Free)

This tool helps product teams triage and prioritize feature requests and product feedback, not IT incident triage. Turn messy requests into a clean, deduplicated, prioritized backlog with decisions, rationale, and customer follow-up messages.

  • Import by paste (one per line), CSV upload, or raw notes. Auto-dedupe and clustering (TF-IDF + cosine) suggest similar requests with shared keywords; merge into canonical requests and preserve evidence.
  • Decisions: Merge, Needs info (with follow-up questions), Accept into Discovery, Park, Reject. Prioritization: reach, impact, confidence, effort, segment/value weight; weighted score. Close-the-loop messages (3 templates × 3 tones).
  • Export CSV, Markdown (stakeholder summary, next actions), JSON, RICE-ready CSV. Tag manager, themes. Share URL. Three examples. No login.

No login. Autosave in browser. Shareable URL. Clear data when you want.

Import:|||CSV template
Load example:
Close-the-loop tone:

Raw requests (0)

    Canonical requests (0)

      Triage details

      Select a canonical request.

      How it works

      1. Import requests by pasting (one per line), uploading CSV (text, source_type, submitted_by, segment, account, plan, mrr_band), or pasting raw notes (support tickets, interview quotes). Toggle auto-dedupe and run triage to cluster similar requests using text similarity (TF-IDF + cosine).
      2. Review clusters and canonical requests in the middle pane. Select a request to set decision (Merge, Needs info, Discovery, Park, Reject), adjust score inputs (impact, confidence, effort), and copy close-the-loop messages. Tag manager and themes help group for stakeholder summaries.
      3. Export CSV (canonical list with score and decision), Markdown (stakeholder summary, top themes, close-the-loop pack, next actions), JSON (full state), or RICE-ready CSV for the RICE calculator. Share URL to restore state. Three loadable examples. No login.

      The 4 steps of triage: capture, dedupe, classify, decide

      Capture: Get feedback from support, sales, interviews, and the portal; track whether it was submitted by the customer or by a teammate (proxy). Dedupe: Use text similarity to group near-duplicates; merge into a canonical request and keep original texts as evidence. Classify: Tag and theme; label type (feature, bug, UX, pricing) and problem vs solution. Decide: Assign Merge, Needs info, Discovery, Park, or Reject with a reason and next step; generate close-the-loop messages.

      Why deduping matters

      Without dedupe you can end up with 50 copies of the same request (e.g. "add dark mode") that look like 50 separate items. Clustering and merging reduce noise, surface true volume, and keep evidence so you can quote customers. The tool shows "why similar" via shared keywords so merges are explainable.

      How to weight requests by segment and value (without lying to yourself)

      Use account, plan, MRR band, or segment so high-value customers aren't drowned out by volume. Proxy submissions (sales or CS on behalf of a customer) should be linked to the account and weighted. The weighted score can include a value weight (e.g. MRR band). Don't over-weight a single loud account; balance reach and value so prioritization stays defensible.

      How to close the loop with customers

      Send status updates so requesters know their feedback was heard. Use three message types: Acknowledged (we received it), In discovery (we're evaluating it), and Shipped/Not doing (decision made). Pick one tone (friendly, concise, enterprise) and use it consistently. The tool generates messages per request so you can copy and send.

      Pro tips

      • Capture source (support, sales, interview, portal) and submitted_by (customer vs internal) so you can weight proxy votes and segment later.
      • Run dedupe before scoring: merge near-duplicates into one canonical request and keep original texts as evidence.
      • Use 'Needs info' when the request lacks who, when, or desired outcome; send 3–6 follow-up questions instead of guessing.
      • Weight by segment and value (e.g. MRR band, plan tier) so high-value accounts aren't drowned out by volume.
      • Tag and theme after clustering: turn top keywords into 3–8 themes for stakeholder summaries.
      • Close the loop with three message types: Acknowledged, In discovery, Shipped/Not doing. Pick one tone (friendly, concise, enterprise) and stay consistent.
      • If confidence is low but impact is high, choose 'Accept into Discovery' and add a next step: 'Run 2–3 interviews to validate.'
      • Document product constraints (e.g. no custom integrations this year) so the engine can suggest Reject with rationale.
      • Export RICE-ready CSV to feed /tools/rice-score-calculator for roadmap prioritization.
      • Review parked and rejected items regularly; communicate decisions so requesters don't feel ignored.

      Common mistakes (symptom, cause, fix)

      Symptom: Treating every request as unique.

      Cause: No dedupe or clustering, so the backlog is a long list of similar items.

      Fix: Run similarity (e.g. TF-IDF + cosine); merge into canonical requests and keep evidence; show 'why similar' keywords.

      Symptom: Prioritizing by volume only.

      Cause: Loudest segment wins even when low value.

      Fix: Add segment and value weight (MRR band, plan); use weighted score so high-value accounts are visible.

      Symptom: No follow-up when context is missing.

      Cause: Requests like 'Add X' get scheduled without knowing who, when, or why.

      Fix: Set decision to Needs info; generate 3–6 follow-up questions and send before prioritizing.

      Symptom: Never closing the loop.

      Cause: Requesters never hear back, so they repeat or churn.

      Fix: Use three message templates (Acknowledged, In discovery, Shipped/Not doing) and send at least one per request.

      Symptom: Solution without problem.

      Cause: Request describes a feature, not the pain or trigger.

      Fix: Classify as Needs info; ask for problem, workaround, and success criteria before discovery.

      Symptom: Ignoring proxy votes.

      Cause: Sales or CS submit on behalf of customers but get no weight.

      Fix: Track submitted_by (customer vs internal); weight proxy submissions and link to account/segment.

      Symptom: No themes for stakeholders.

      Cause: Backlog is a flat list; execs can't see patterns.

      Fix: Build 3–8 themes from tags; summarize top themes and top 10 requests with decisions in stakeholder export.

      Symptom: Rejecting without rationale.

      Cause: Requesters don't know why their idea was declined.

      Fix: Document product constraints; use Reject with a reason and next_step; include in close-the-loop message.

      FAQ

      Is this for product feedback triage or IT incident triage?

      This tool is for product teams triaging feature requests and product feedback, not IT incident or support ticket triage. You get dedupe, clustering, prioritization (reach, impact, confidence, effort), decisions (merge, needs info, discovery, park, reject), and close-the-loop messages for customers.

      What are the three input methods?

      You can paste requests (one per line), upload a CSV with columns such as text, source_type, submitted_by, segment, account, plan, mrr_band, and source_link, or paste raw notes (support tickets, interview quotes) and the tool will split them into requests. All run client-side; no login required.

      How does dedupe and clustering work?

      The tool uses text similarity (TF-IDF and cosine similarity) to find near-duplicate requests. It suggests clusters with shared keywords so you can see why items are similar. You can merge into a canonical request; original texts are kept as evidence. Auto-dedupe can run on import, or you can adjust the threshold and run manually.

      What is a proxy request?

      A request submitted by an internal teammate (e.g. sales or CS) on behalf of a customer. The tool tracks submitted_by (customer vs internal) so you can weight proxy votes and link them to account, plan, or segment. This avoids losing high-value feedback that comes through sales instead of the portal.

      What decisions can the engine suggest?

      Merge (for duplicates), Needs info (missing context; generates follow-up questions), Accept into Discovery, Park (later), and Reject (e.g. conflicts with product constraints). Each has a reason and next step. You can override any decision in the triage details panel before exporting or sharing.

      What is the weighted score formula?

      The default formula is (Impact × Confidence × Reach weight) / Effort, with optional value weight. Impact, confidence, and effort are 1–5. Reach weight is often the requester count (capped). This is not RICE unless you map reach to RICE reach; you can export a RICE-ready CSV for the RICE calculator tool.

      What are close-the-loop messages?

      Three templates per request: Acknowledged (we received it), In discovery (we're looking into it), and Shipped/Not doing (decision made). You pick a tone (friendly, concise, enterprise). The tool fills in the request title so you can copy and send to customers.

      Can I export to the RICE calculator?

      Yes. Export the optional RICE-ready CSV with id, title, reach, impact, confidence, effort, score, and decision. Import that into the RICE Score Calculator tool to refine scores or combine with other roadmap items. The triage tool does not implement full RICE scoring unless you align the fields yourself.

      How do themes and tags work?

      Tags can be suggested from keywords; you confirm or rename. Themes group tags (e.g. 3–8 themes like First-time experience, Pricing, Engagement). Use them in the stakeholder summary so execs see patterns. You can merge or rename tags in the tag manager.

      Is the Feature Request Triage Assistant free?

      Yes. The tool is free, runs in your browser, and requires no login. You get paste, CSV, and raw-notes import; dedupe and clustering; decisions and scoring; close-the-loop messages; CSV, Markdown, and JSON export; and a shareable URL. Autosave and three loadable examples included.

      Learn more with CraftUp

      Courses, blog, and glossary for product and execution.

      Product DiscoveryMaster Problem ValidationProduct feedbackRoadmap templateValue propositionKPI

      From feedback to roadmap

      Use CraftUp tools and courses to triage feedback, prioritize, and close the loop with customers.

      Try the appExplore courses

      Freshness

      Last updated: 2026-03-06

      • 2026-03-06: Launched Feature Request Triage Assistant: paste, CSV, raw-notes import; dedupe and clustering (TF-IDF + cosine); decisions and weighted score.
      • 2026-03-06: Close-the-loop messages (3 templates × 3 tones); tag manager; themes; CSV/MD/JSON and RICE-ready export; shareable URL; 3 examples.
      • 2026-03-06: Proxy requests (customer vs internal); Needs info with follow-up questions; product constraints for Reject; stakeholder summary and next actions.