Quick Score — Targeted Batch Scoring for Lead Gen

Quick Score — Targeted Batch Scoring for Lead Gen

This skill scores a user-provided list of companies using the full RemidiWorks v4 methodology (Binary Fact Sheet, 28 sub-dimensions, constraint rules, cross-checks) and produces outreach-ready summaries comparing each company's scores to the database averages.

How This Differs From Score-Day and Rapid-Scan

The scoring methodology is identical to score-day. The difference is the input source (user-provided list vs. Target_List_500) and the output format (outreach summaries on top of the standard DB update).

Pipeline

Phase 0 → Read rubric & config Phase 1 → Receive and validate the target list Phase 2 → Research (parallelized, same as score-day) Phase 3 → Score using full v4 framework Phase 4 → Write evidence files Phase 5 → Update master database Phase 6 → Generate outreach summaries (the extra layer)

Phases 0–5 follow the score-day skill exactly. Phase 6 is unique to quick-score.

Phase 0: Read the Rubric

Before doing anything else, read the canonical scoring rubric:

RemidiWorks/01_Methodology/Config/batch_scoring_agent_prompt_v3.md

This contains the v4 category architecture, Binary Fact Sheet (F1–F30), constraint rules, scoring rubrics for all 28 sub-dimensions, anti-uniformity checks, and cross-check verification. Do NOT proceed to research until this file has been read.

Phase 1: Receive the Target List

The user will provide companies in one of these formats:

• Pasted list of company names or URLs in the chat
• Excel/CSV file with company names and URLs
• Reference to an existing file (e.g., "score the next 5 from Target_List_500")

For each company, you need at minimum: company name and URL. If the user provides only names, look up the URLs. If they provide only URLs, determine the company names from the websites.

Validate before proceeding:

1. Confirm the list with the user: "I'll score these [N] companies: [list]. Correct?"
2. Check if any are already in the master database (search by URL/domain in Rapid_Scan_Master_Database.xlsx). If found, flag it: "FYI, [company] was already scored on [date] with overall=[X]. Want me to rescore or skip?"
3. Confirm batch size is manageable (recommended: 3–10 for quick-score; if >10, suggest splitting or using score-day)

Phases 2–5: Score-Day Methodology

Follow the score-day skill for these phases. The process is identical:

Phase 2: Research (Parallelized)

• Launch parallel research agents (1 agent per 5 companies)
• Each agent uses Chrome browser + WebSearch to gather homepage, pricing, case studies, validation data, funding data
• Buyer-platform identification: For each company, identify the buyer persona and their preferred validation source using the lookup table in batch_scoring_agent_prompt_v3.md. Search the buyer-relevant platform first, then G2 for the master database.
• Collect study data for every company (buyer-relevant platform data, G2 rating, review count, category, badge status, funding stage)
• Follow Chrome-first approach for company websites; WebSearch for external platforms

Phase 3: Score Using v4 Framework

• Complete Binary Fact Sheet (F1–F30) for each company
• Score all 28 sub-dimensions using constraint rules
• Run anti-uniformity check and cross-check verification
• Calculate category scores and overall using Python helpers
• Assign tiers

Phase 4: Write Evidence Files

• Save assessment_scores.json, scoring_verification.md, research_notes.md per company
• Location: RemidiWorks/companies/{slug}/assessments/{date}_rapid_scan/

Phase 5: Update Master Database

• Add rows to Rapid_Scan_Master_Database.xlsx (all columns per the canon column map (currently 49 columns))
• Write sub-dimension scores to the Sub-Dimension Detail tab
• Recalculate percentile ranks for all comparable companies
• Calculate Winner Cohort for new companies using the B+ winner definition (see scoring_canon.md section 3):
  • Primary gate: G2 Leader badge + ≥200 G2 reviews → Winner (Y)
  • Safety valve override: If NOT Leader, but has 1000+ G2 reviews AND 4.5+ G2 rating AND badge_status != "None" → Winner (Y)
  • Field (N): Has review platform presence but meets neither condition
  • IND: No review platform presence
• DO NOT update Target_List_500.xlsx unless the companies came from there
• Scan History Tab (MANDATORY): After writing to the Master Database sheet, append a row for EVERY scored company to the Scan History tab with: Scan Date (today), Company Name, URL, Score Version ("v4"), all v4 scores, Source Skill = "quick-score", Notes, PE Sponsor, Fund Name. This is append-only — never overwrite existing Scan History rows. If rescoring an existing company, update the Master Database row AND append a new Scan History row.
• Save and commit

Phase 6: Generate Outreach Summaries

This is the unique deliverable. After scoring is complete, produce a one-page outreach summary for each company plus a batch overview.

Database Averages (from references/db_averages.json)

These are a snapshot from references/db_averages.json (last updated 2026-04-11, n=523 comparable companies). Always recompute from the live database after adding new companies — these numbers go stale with every batch.

Note: PP average is among the ~72% of companies with scored PP (n=374). CP average based on n=402 (192 legacy records missing CP). For companies with PP=N/A, show "N/A" in the table and note it.

Important: After adding the new companies to the database, recompute the averages from the updated database for accuracy. The pre-computed numbers above are a starting reference.

Per-Company Outreach Summary

For each scored company, output this format:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
QUICK SCORE: [Company Name]
URL: [url]  |  Vertical: [vertical]
Scanned: [date]  |  Database: [N] B2B SaaS companies
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

                         Company    DB Avg    Delta
─────────────────────────────────────────────────────
OVERALL                    [XX]       55     [+/-XX]
─────────────────────────────────────────────────────
Messaging & Positioning    [XX]       71     [+/-XX]
Pricing & Packaging        [XX]       62     [+/-XX]
Buyer Experience           [XX]       51     [+/-XX]
Trust & Credibility        [XX]       56     [+/-XX]
Competitive Position       [XX]       54     [+/-XX]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Tier: [T0/T1/T2/T3] — [Tier Name]
Biggest Gap: [Dimension] at [XX] vs. avg [XX] — [1-sentence observation]
Biggest Strength: [Dimension] at [XX] vs. avg [XX] — [1-sentence observation]

── LINKEDIN DM ──────────────────────────────────────

I just ran a commercial health assessment on [Company Name] —
you scored [XX] against a database average of 55 across [N]
B2B SaaS companies. Your biggest gap is [dimension] at [XX]
vs. the average of [avg]. Happy to walk you through the
findings if useful.

── EMAIL ─────────────────────────────────────────────

Subject: [Company Name] scores [XX] on commercial health —
here's where the gaps are

[First name] — I evaluate B2B SaaS companies on 5 dimensions
of commercial health (messaging, pricing, buyer experience,
trust, competitive position) using a database of [N] companies.

[Company Name] scored [XX] overall. Your strongest area is
[dimension] at [XX] (vs. avg [XX]). Your biggest gap is
[dimension] at [XX] (vs. avg [XX]).

Worth a 15-minute call to walk through the details?

── INTERNAL NOTES ────────────────────────────────────

MP ([XX]): [One sentence — key observation]
PP ([XX]): [One sentence — key observation]
BE ([XX]): [One sentence — key observation]
TC ([XX]): [One sentence — key observation]
CP ([XX]): [One sentence — key observation]

Top 3 Opportunities:
1. [From assessment_scores.json top_opportunities]
2. [...]
3. [...]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

For PP = N/A companies:

Pricing & Packaging        N/A        62       —

Add note: "PP excluded — no public pricing detected. Overall calculated from 4 dimensions."

Batch Overview

After all per-company summaries, produce a batch overview:

═══════════════════════════════════════════════════
QUICK SCORE BATCH SUMMARY
Scored: [N] companies  |  Date: [date]
═══════════════════════════════════════════════════

Company                Overall  Tier   Biggest Gap
────────────────────────────────────────────────────
[Company 1]              [XX]   [TX]   [Dimension] ([XX] vs [avg])
[Company 2]              [XX]   [TX]   [Dimension] ([XX] vs [avg])
[Company 3]              [XX]   [TX]   [Dimension] ([XX] vs [avg])
...
────────────────────────────────────────────────────
Batch Average:           [XX]          DB Average: 55

Tier Breakdown: T3: [n]  T2: [n]  T1: [n]  T0: [n]
Database updated: [total] companies now scored
═══════════════════════════════════════════════════

Output Files

Save the outreach summaries to a single file:

RemidiWorks/companies/_outreach/quick_score_batch_{date}.txt

This keeps outreach copy separate from the evidence files (which go in each company's folder). Present the file to the user with a computer:// link.

Outreach Copy Guidelines

The LinkedIn DM and email drafts are templates the user will customize. Keep them:

• Direct. Lead with the score. The number is the hook.
• Specific. Name the biggest gap and its score vs. average. Generic "I have insights" messages get ignored.
• Short. LinkedIn DM: under 300 characters. Email: 3–4 sentences max in the body.
• Not salesy. "Happy to walk through the findings" — not "Book a call to learn how we can help."
• Honest. If the company scored well, say so. Lead with their strength and frame the gap as "even stronger if." Don't manufacture urgency where none exists.
• Buyer-aware. Do not reference G2 reviews or any review platform in outreach if that platform isn't where the company's buyer evaluates vendors. Calling out thin G2 presence to a company selling to retail merchandisers or healthcare administrators signals that you don't understand their market.

For companies scoring above the database average (55+), lead with their strength and position the gap as an optimization opportunity. For companies below average, lead with the gap — that's where the urgency is.

Edge Cases

• Company already in database: Ask user — rescore or skip? If rescoring, note "Rescore" in the Notes column, update the existing Master Database row (don't duplicate), and append a new row to the Scan History tab.
• Non-B2B SaaS company in the list: Flag it, ask the user. If they want it scored anyway, proceed with a Framework Fit note.
• Fewer than 3 companies: Still works — no minimum batch size. Just less efficient than a single rapid-scan.
• More than 10 companies: Suggest splitting into batches or using score-day. Quick-score is designed for targeted outreach lists, not database-building runs.
• User provides just names, no URLs: Look up the URLs. Confirm with user before proceeding.
• PP = N/A: Common (~45% of database). Handle per the PP N/A rule in v4 framework.
• Chrome timing out: Fall back to WebFetch + WebSearch. Flag LC-2 for visual element scores.

QC Gate Protocol — Scoring Model Integrity Checks

Add this section to every scoring skill (score-day, rapid-scan, quick-score, report-card, portfolio-scoring) as a mandatory post-scoring verification step. No scores should be written to the Master Database or presented to a client until ALL gates pass.

Gate 1: Formula Reproduction (run EVERY batch)

After calculating scores for any company, verify the math reproduces:

# TC verification: sum of 8 sub-scores / 40 * 100
tc_subs = [scores['4A'], scores['4B'], scores['4C'], scores['4E'],
           scores['4F'], scores['4G'], scores['4H'], scores['4I']]
# NOTE: 4D (Review Platforms) is EXCLUDED from TC. Collected as enrichment only.
tc_calc = round(sum(tc_subs) / 40 * 100)
assert abs(tc_calc - stored_tc) <= 1, f"TC mismatch: calc={tc_calc}, stored={stored_tc}"

# Overall verification: canonical v4 weights
if pp_na:
    overall_calc = round(mp*(25/75) + be*(15/75) + tc*(20/75) + cp*(15/75))