Agency Prospect Scoring System: Rank Local Leads by Website Weakness

By Cameron Kirdzik — Founder @WebHunt.ai

September 29, 2026 · 9 min read

Why scoring beats volume in the AI era

AI build tools collapsed cost and time. A 3–5 page brochure site that once took weeks now takes hours with Replit, Lovable, Base44, v0, Bolt, Framer AI, Cursor, Windsurf, Codex Sites, Claude Design, or Claude Code. Your ceiling isn’t fulfillment; it’s qualification. The operator who processes 50+ leads a week and prioritizes the right 10 will out‑earn the designer waiting on two referrals.

Without a score, we burn hours demoing for low‑upside leads while hot, broken‑site prospects age out. A scoring system stacks objective weakness signals into one number so your pipeline sorts by revenue‑likelihood, not recency or gut feel.

The AI‑era pitch advantage (same‑day draft, low‑risk pilot) is real—but it’s a time investment. Deploy it only on 80+ point leads. Tie score bands to actions:

• 80–100 = immediate personalized demo
• 50–79 = audited teardown + ROI case
• 0–49 = nurture with lighter touches

Key takeaway: Speed is your edge, but score determines where to spend it.

The signals to score: what actually predicts readiness to buy

Score signals that correlate with money‑on‑the‑table and are fast to verify:

• Core site quality
  • Missing site (or FB‑only)
  • Non‑responsive/mobile failures; mobile‑friendliness is a ranking input in Search¹
  • Performance: Core Web Vitals; LCP > 4s on mobile = severe²
  • Obvious template rot or 5+ year‑old visual patterns
• Conversion blockers
  • No clickable phone on mobile
  • No contact/booking form
  • No above‑the‑fold CTA
  • Broken links
• Freshness
  • Last meaningful update > 24 months
  • Stale copyright footer
  • Old promos still live (e.g., 2021 COVID banner)
• Local demand and review momentum
  • Very low review count relative to top‑3 competitors
  • No new reviews in 90 days
  • Google Business Profile (GBP) unclaimed or thin
• Trust/technical
  • HTTPS missing or misconfigured; modern browsers and users expect HTTPS and Google recommends it¹
  • Missing LocalBusiness schema; missing analytics/tracking
• Competitive gap proxies
  • Authority/traffic well below local peers suggests room to leapfrog with fundamentals
• Operational fit
  • High‑LTV, phone‑first trades (HVAC, roofing, pest control) can ROI a redesign with one booking—weight them higher

By the numbers: As mobile page load increases, bounce rates rise sharply; speed isn’t cosmetic, it’s cash².

Build your agency prospect scoring system: a 100‑point rubric

Use a 100‑point composite score so assistants and automations can score consistently. Award points when the problem exists (the worse the site, the higher the score). Cap at 100.

• Status & security (25 pts)
  • No site or FB‑only: +25
  • Has site but no HTTPS: +15
  • HTTPS misconfigured/mixed content: +10
• Mobile & speed (20 pts)
  • Fails responsive layout or tap targets: +10
  • LCP > 4s on mobile: +10
• Conversion readiness (15 pts)
  • No clickable phone on mobile: +5
  • No form/booking: +5
  • No above‑the‑fold CTA: +5
• Freshness (10 pts)
  • No updates > 24 months: +6
  • Outdated promos/copyright: +4
• Review momentum (10 pts)
  • Reviews < 30% of category median: +6
  • No new reviews in 90 days: +4
• GBP/Local presence (5 pts)
  • GBP unclaimed/incomplete: +5
• Technical/SEO basics (10 pts)
  • No analytics tag: +3
  • No LocalBusiness schema: +3
  • Title/meta obvious defaults/duplicates: +4
• Competitive gap (5 pts)
  • Domain authority or traffic below bottom quartile locally: +5
• Trade multiplier (bonus up to +10)
  • High‑LTV phone‑first (HVAC, roofing, pest): +10
  • Mid‑LTV (dental, legal): +5
  • Low‑LTV retail/food: +0

Always document scoring notes per lead so your outreach references exact issues.

Make inputs repeatable (fast heuristics that scale)

• Competitive gap: Pull DA with MozBar or Ahrefs Webmaster Tools. Sample 20 local competitors in the same city/category. Define “bottom quartile” as DA below the 25th percentile of that sample.
• Review momentum: In GBP, collect review counts for the top 3 ranking competitors for the main keyword (city + category). Compute the median. If your prospect has < 30% of that median, flag it.
• Freshness: Check the homepage in the Wayback Machine once per quarter and visually diff for layout/content changes; or grab the most recent blog/news timestamp. If no visible updates in > 24 months, flag it.

Data collection: fast ways to populate your rubric at scale

You don’t need a heavyweight crawler to start. A quick‑check stack covers most signals:

• Lighthouse mobile test for LCP/CLS (Core Web Vitals)
• Chrome DevTools mobile emulator for tap targets and click‑to‑call
• Flag click‑to‑call by inspecting anchor tags for tel: on mobile view
• View Source (or Tag Assistant) for analytics and schema
• Wayback Machine for visual age and last update
• GBP check for claim status, review counts, and recency; compare to top‑3 in the same city/category

Automation options:

• Use site audit APIs for HTTPS, title/meta, and speed flags; fetch into Google Sheets via Apps Script.
• Auto‑screenshot homepages (desktop and mobile) and have an LLM summarize breakages into a “Notes” field; reserve human review for 60+ scores.
• Enrich owners early so high‑score leads get a fast call instead of a generic email. If you’d rather not build the list by hand, WebHunt.ai surfaces local businesses pre‑scored for website‑weakness and demand signals (HTTPS, mobile, speed, freshness, review momentum) so your spreadsheet starts with likely buyers.

For faster dials on A‑tier, WebHunt.ai can enrich owner contacts (name, direct phone with line‑type, and email) so you start with the right person.

Your working spreadsheet: columns, CSV header, and formulas

Copy/paste this CSV header row into Sheets to create the exact columns:

Business,URL,City,Category,NoSite_FBOnly,NoHTTPS,HTTPS_Mixed,MobileFail,LCPgt4s,NoClickPhone,NoFormBooking,NoAboveFoldCTA,Stale24mo,OutdatedPromo,LowReviewsPct,NoNewReviews90d,GBP_Unclaimed,NoAnalytics,NoLocalSchema,MetaDefaults,LowDA_Traffic,TradeBonus,Score,Tier,OwnerName,DirectPhone,Email,Notes,DateScored

Each signal column is 1/0; TradeBonus is 0/5/10. Score sums weighted points and caps at 100.

• Simple named‑field version (optional):

=MIN(100, (25NoSite_FBOnly + 15NoHTTPS + 10HTTPS_Mixed + 10MobileFail + 10LCPgt4s + 5NoClickPhone + 5NoFormBooking + 5NoAboveFoldCTA + 6Stale24mo + 4OutdatedPromo + 6LowReviewsPct + 4NoNewReviews90d + 5GBP_Unclaimed + 3NoAnalytics + 3NoLocalSchema + 4MetaDefaults + 5*LowDA_Traffic + TradeBonus))

• Copy/paste‑proof SUMPRODUCT version using column letters (assumes binary signals live in E:T and TradeBonus in U; place this in row 2 of Score column (V2)):

=MIN(100, SUMPRODUCT({25,15,10,10,10,5,5,5,6,4,6,4,5,3,3,4,5}, E2:T2) + U2)

Auto‑assign tier (W2):

=IF(V2>=80, "A (Hot)", IF(V2>=50, "B (Workable)", "C (Nurture)"))

Filter views:

• A‑tier: sort by Score desc
• B‑tier: group by Category/City for batch teardowns
• C‑tier: hide or bulk‑nurture

Create a weekly snapshot (copy‑paste values in a new tab) to measure conversion by tier over time. If you pull leads from WebHunt.ai, you can export with built‑in weakness signals and AI opportunity briefs, then map them directly to these columns to skip manual data entry.

From score to script: change the pitch by tier

Tie the score to your talk track and your time.

• A‑tier (80–100)

  • Lead with urgency and a 24–48h live draft promise.
  • Reference 2–3 specific breakages (e.g., click‑to‑call missing, LCP 5.2s, HTTPS missing).
  • Offer a low‑risk pilot or “try before you commit.”
  • Example opener (live call): “Spotted your site timing out on mobile and the phone number isn’t tappable. I can turn around a mobile‑first draft by tomorrow so you don’t miss weekend calls—want to see it?”
  • Voicemail/SMS variant: “Quick heads‑up: your site’s not tappable on mobile and loads in ~5s. I can send a working draft by tomorrow. Reply YES for the draft link.”

• B‑tier (50–79)

  • Lead with ROI math and a short audit video.
  • Propose a fixed‑price refresh with practical add‑ons.
  • Price anchor: “$400 fast refresh + $200 review engine add‑on; 1–2 day turnaround.”
  • Example opener: “You’re close. Two fixes—mobile tap targets and adding click‑to‑call—plus a simple review ask could move you into the top pack. I recorded a 3‑minute walkthrough with exactly where.”

• C‑tier (0–49)

  • Low‑effort nurture: send a one‑page teardown PDF with annotated screenshots; quarterly check‑in.
  • No custom demo until score improves; set a reminder to re‑score in 90 days.
  • Nurture trigger: +15 if they add 10 reviews in 90 days or claim/fill GBP; re‑route to B‑tier.

Use the score to justify follow‑up pacing:

• A‑tier = call same day + SMS + email with demo link
• B‑tier = email + Loom + call next day
• C‑tier = email only

If you prefer to augment calling, WebHunt.ai offers an AI Voice Agent SDR that calls leads, pitches on your behalf, and books qualified meetings into your Google/Outlook calendar with compliance guardrails. Prefer human voice? The WebHunt.ai human cold‑calling marketplace can work your A‑tier list and land meetings directly on your calendar while you build.

Workflow: route, demo, and book using automation

Daily ritual:

• Pull 30–50 fresh leads, score in bulk.
• Route A‑tier to same‑day call queue; B‑tier to audit‑creation queue; C‑tier to nurture list.
• Attach visual proof (desktop/mobile screenshots) in first outreach so prospects instantly see what’s broken.
• Include two suggested times and a booking link in every message.

Demo efficiently:

• Spin a draft for A‑tier in hours using Replit, Lovable, Base44, v0, Bolt, Framer AI, Cursor, Windsurf, Codex Sites, Claude Design, or Claude Code.
• Paste real business details and photos to lift reply rates. If you want a head start, WebHunt.ai can generate a one‑click website prompt pre‑filled with the business’s real info that you can paste into your builder of choice to produce a credible first draft fast.

Pipeline math:

• Aim for 25 A‑tier conversations/week at ~5% close = 1–2 new projects. With near‑zero fulfillment cost on AI‑built sites, targeted throughput beats broad untargeted blasts.

Calibration: validate and tune your rubric with real outcomes

Run a 4‑week pilot:

• Score 200 leads; tag outcomes (connected, meeting set, closed, lost); calculate close rates by tier.
• If B‑tier closes nearly as well as A‑tier in a niche, adjust weights upward on the signals that differentiated A vs. B (often review momentum or conversion blockers).
• Segment by trade/city. Roofing during storm season may warrant a temporary +5 bonus; restaurants might need lower review weighting compared to booking availability.
• Add a feedback loop: after each close, log which 3 issues the owner cared about most; mirror that language in your templates.
• Re‑score quarterly. Sites improve or decay; your A‑tier list should stay fresh.

If you’re sourcing from WebHunt.ai, use the built‑in deal pipeline to track outcomes by tier and export results for analysis. Their AI opportunity briefs also surface the exact breakages (with screenshots) that owners are likely to react to, which helps with message tuning.

Example scenario for illustration; results vary

• Pull: Source 150 home‑service leads in a single metro; score in a Sheet using the rubric. Outcome pattern you might see: ~10–20 A‑tier, ~40–70 B‑tier, remainder C‑tier.
• Produce: Build 8–12 A‑tier demos within 24 hours using AI site builders; each demo references LCP, click‑to‑call, and HTTPS issues.
• Outreach cadence: Day 1 calls to all A‑tier with demo links; Day 2 audit videos for B‑tier; Day 3 follow‑ups.
• Meetings: Teams commonly report 2–8 meetings booked from this cadence when the list quality is high. Protect demo time for 80+.

Process tip: Document exactly which issues the owner reacted to on each call; use that to refine your opener lines and weightings.

Ready to put this to work?

Score fast, route by tier, demo only for 80+, and tune monthly against close rates. If you want a head start on qualified leads and owner contacts, grab a free account at WebHunt.ai to find local businesses already scored for website weakness and buying likelihood.