Test Ride Landing Page & Campaign Strategy
Overview
During a weekly coaching call on 2026-04-05, Mark Hope demoed a Claude-powered workflow for Crazy Lenny's e-bike store in Madison, WI. Starting from a single objective — drive in-store foot traffic — the session produced a complete HTML landing page, a thank-you page, a full Google Ads campaign strategy, and a Start/Stop/Continue analysis against live campaign data. Total elapsed time: under 45 minutes.
Attendees: Ben San Fratello, Mark Hope
Related meeting: [1]
Client: [2]
Primary Conversion
"Schedule a Test Ride"
Chosen because it is a low-commitment, high-value offer that guarantees a warm lead in the store. Once someone shows up for a test ride, the sales team has a captive, interested prospect.
Supporting CTAs considered but deprioritized: Get Directions, Check Store Hours, Reserve Your E-Bike.
Workflow: Step by Step
Step 1 — Define the Objective
Before touching a landing page or campaign, identify the single most important conversion for the business.
For Crazy Lenny's: they don't sell online. The only thing that matters is foot traffic. Every campaign and every landing page should funnel toward that.
"Begin with the end." — Stephen Covey, cited by Ben during the call
Step 2 — Generate the Landing Page (HTML)
Tool: Claude (preferred over ChatGPT for web browsing and HTML generation)
Prompt pattern:
1. Provide the client URL so Claude can browse the site and gather real context.
2. Describe the business and traffic source.
3. Ask Claude to recommend the primary CTA (optional — can be specified directly).
4. Confirm the page structure before building.
5. Say explicitly: "Build the page using HTML."
Resulting page structure included:
- Hero with CTA button ("Schedule Your Free Test Ride")
- Friction reducers ("Takes 30 seconds to book, no purchase required")
- Social proof — real testimonials pulled from the Crazy Lenny's website and Yelp
- Selection preview with bike category links
- Urgency/scarcity block ("This Weekend's Test Ride Schedule — 3 slots remaining")
- How It Works (Book → Meet Expert → Hit the Road)
- FAQ
- Sticky mobile footer CTA
Refinements made in-session:
- "Use only real testimonials from the website." — Claude pulled reviews mentioning Evan (store manager) by name
- "Move the 'What Madison Riders Say' section above the 'Test Ride Our Largest Collection' section." — Claude rewrote the relevant HTML block
- "Make the CTA button purple." — Claude updated the CSS including hover state
Key insight: Specify everything you want in the initial prompt. Post-hoc edits work but are slower because Claude must re-parse the full HTML.
Step 3 — Generate the Thank-You Page
Prompt: "Now create a thank you page for the test ride confirmation."
Resulting page included:
- Appointment confirmation block (date, time, duration, confirmation number)
- "What Happens Next" checklist (check email → arrive → meet expert → test ride)
- Map embed placeholder
- "What to Bring" checklist (driver's license, comfortable clothing)
- Helpful resource links
- Professional footer
Step 4 — Generate the Google Ads Campaign Strategy
Prompt: "Given this landing page with this CTA, what Google Ads campaigns should we run to send traffic to it?"
Output included:
| Campaign | Focus | Budget % |
|---|---|---|
| 1. Branded | Crazy Lenny's brand terms | ~20% |
| 2. E-Bike Purchase Intent | "e-bikes for sale Madison," "best e-bike dealer" | ~25% |
| 3. Test Ride Focus | Searches for test rides | ~40% |
| 4. E-Bike Categories | Specific bike types | ~10% |
| 5. Competitor Conquesting | Competitor brand terms | ~5% |
Also generated:
- 15 headlines + descriptions (ready to paste into Google Ads)
- Ad extensions
- Negative keyword list
- URL structure and UTM parameter recommendations
Step 5 — Start/Stop/Continue Analysis
Prompt: "Compare these recommendations to the current campaign data [uploaded Excel export from Google Ads] and tell us what to Start, Stop, and Continue."
Claude parsed the Excel file, ran analysis, and returned:
Continue (scale these):
- Local Store Visits Push — CPA: $1.86, 224 conversions. Best performer. Increase budget aggressively.
- [High-performing branded campaign] — 4.36 CPA, 19% conversion rate. Keep and expand.
Stop (pause immediately):
- Retargeting Display Campaign — 565 clicks, 0 conversions. No justification to continue spend.
Start (new campaigns based on coverage gaps):
- Test Ride Focus campaign — not currently running; directly supports the new landing page
- Competitor Conquesting campaign — not currently running; low-cost brand awareness play
Revised budget recommendation: Increase from ~$70/day to ~$120/day, reallocated per the table above.
Handoff Plan
| Artifact | Recipient | Action |
|---|---|---|
| Landing page HTML + thank-you page HTML | Melissa (design) / Gilbert (dev) | Build and deploy |
| Campaign strategy + ad copy | Gilbert | Implement new structure in Google Ads |
| Start/Stop/Continue analysis | Carly (account manager) | Review with client; get buy-in on budget increase |
| Landing page preview link (Claude artifact URL) | Carly | Share with client for feedback before build |
Recommended handoff format: Paste HTML into a Google Doc (not raw into Slack — it's too long). Share the Claude artifact publish link for the visual preview.
Action Items
- [ ] Send Crazy Lenny's landing page + thank-you page artifacts and campaign recommendations to Carly and Gilbert; cc Melissa; request build — @Ben San Fratello
- [ ] Draft Exterior Renovations landing page (CTA: "Get a Free Estimate") in Claude using same workflow; send to Melissa for build — @Ben San Fratello
Prompting Tips Captured
- Give Claude the URL first. It will browse the site (and Yelp, ZoomInfo, etc.) to gather real context. Optionally follow up with
"Tell me about this business"to verify it understood before proceeding. - Specify data sources upfront. Say
"use only real testimonials"in the original prompt, not after the fact. - Upload brand guides. If a brand guide exists, attach it so Claude uses correct fonts, colors, and tone from the start.
- Iterate in-platform. Give direct, specific commands to reshape the page without leaving Claude.
- Structure analysis requests. Use "Start, Stop, Continue" framing to prevent Claude from going off on tangents.
- HTML > other formats. Requesting HTML produces a functional wireframe that designers can paste directly into Figma or a browser. Other formats produce visual mockups that are harder to work with downstream.
AI Tool Notes
| Tool | Best For |
|---|---|
| Claude | HTML generation, web browsing, structured analysis, campaign strategy |
| ChatGPT | Conversational Q&A; weaker at code and web research |
| Perplexity | Fact-finding with citations |
| Gemini | Google ecosystem tasks (Gmail, Google Ads context) |
See also: [3]
Related
- [2]
- [4]
- [5]
- [6]
- [1]