Founding internal case study · Aggregate data only · Data through September 2, 2026
How The Branding Iron documented AI-assisted acquisition after a website overhaul.
The Branding Iron launched a substantially rebuilt website on July 19, 2026. During the matched post-launch period, tracked AI-assistant traffic grew faster than overall traffic. A later customer-level audit connected documented AI-assisted acquisition to paid customer relationships and collected Stripe revenue.
Executive finding
Across the Stripe export reviewed through September 2, The Branding Iron collected $42,160.50 through 84 paid charges associated with 70 unique paying customers. Customer-level reconciliation established affirmative AI-acquisition evidence for 35 customers. Their 43 charges accounted for at least $20,710.50, or 49.1%, of collected revenue reviewed.
Total sessions
318 → 447+40.6%AI-assistant sessions
12 → 53+341.7%AI share of sessions
3.8% → 11.9%approximately 3.1×Confirmed AI-acquired customers
35 of 7050.0%Associated paid charges
43 of 8451.2%Associated collected revenue
$20,710.5049.1% lower boundStarting point
The old site was essentially a four-path brochure, with approximately 96.8% of pageviews in the observed inventory concentrated on the homepage. The completed new-site inventory was described as 66 pages, with 40 post-launch paths receiving traffic through the cutoff.
What changed
- Narrower positioning and answer-first content
- Transparent pricing and decision-stage scope
- Industry relevance and verifiable authority
- Technical legibility and internal linking
- Conversion architecture and attribution
Matched-period traffic
Equal adjacent 46-day periods were used: June 3 through July 18, and July 19 through September 2, 2026.
Source: GA4. Equal periods remove duration as an explanation, but do not control for seasonality, demand, platform adoption, other marketing, or firm activity.
Consultation evidence
Of 70 non-test consultation notifications, 42 contained a ChatGPT signal somewhere in captured attribution fields. Thirty-two recorded ChatGPT as the first channel. These are notifications, not deduplicated clients or conversion rates.
Paying-customer reconciliation
Half of 70 unique paying customers had affirmative AI-acquisition evidence. Those relationships accounted for 43 of 84 paid charges and at least $20,710.50 of $42,160.50 collected.
49.1% tied to customers with affirmative evidence · 50.9% without affirmative evidence established
Attribution maturity changed the measured result.
| Evidence level | Documented amount |
|---|---|
| Technical clearance-stage tracking | $1,830 |
| Plus clearance-stage self-report | $4,495 |
| Initial cross-stage reconciliation | $5,440 |
| Full paying-customer reconciliation | $20,710.50 |
- Technical
- $1,830
- Plus self-report
- $4,495
- Cross-stage
- $5,440
- Full customer
- $20,710.50
The increase resulted from broader evidence coverage and changing the unit of analysis from an immediate conversion record to the acquired paying-customer relationship. Better measurement identified revenue already collected; it did not create it.
Acquisition source and conversion source answer different questions.
Prospect asks an assistant.
Direct, Google, or referral.
Attorney follow-up or email.
Stripe or direct link.
What the evidence supports
The timing and attribution records support the conclusion that AI assistants became a meaningful referral source after the overhaul. The evidence shows a commercially meaningful association between documented AI-assisted acquisition and collected revenue.
What it does not prove
- The overhaul caused a particular referral
- Any page, schema property, crawler rule, or file caused a recommendation
- $20,710.50 is an exact estimate of total AI influence
- The remaining revenue was acquired without AI
- Another firm should expect comparable results
Implementation lessons
Make expertise legible
Focused positioning, hiring-stage answers, transparent scope, and verifiable credentials reduce ambiguity.
Measure the relationship
Acquisition source and conversion source should remain separate across consultation, intake, and payment.
Do not isolate one tactic
Content, authority, technical access, conversion, and attribution changed together. No single cause was tested.
Download the complete 42-page report.
Accessible PDF · 72 pages · approximately 10.8 MiB · Original 42-page design plus tagged text transcript · No marketing opt-in required
If the download does not begin, open the PDF in a new tab.
Want future CounselBeacon research?
Subscribe for occasional findings about AI-assisted legal referrals and law-firm measurement.