Last month the foundation went live. This cycle we shipped the piece that connects it to the outside world: on July 29 we fixed your map pin and linked all six of your new directory and social profiles directly to your business identity in the site's code (verified in the raw code AI systems actually read). Twenty-seven days later, the honest picture is exactly what we'd expect at day 27: the early signals are moving, the headline number is not yet — and we won't claim it is until the pre-agreed week-12 measurement in late October.
| The question | How we measure it | What happened | Answer |
|---|---|---|---|
| Is your identity + profile network AI-readable? | Fetching the raw code exactly as AI crawlers do | Upgrade live Jul 29: map pin fixed, 6 profiles linked — re-verified this cycle | ✓ shipped & verified |
| Do AI assistants recommend Joybody more? | Mentions in 200 real patient questions, 4 platforms | 8.5% → 8.5% — flat, as expected this soon after the profile work | ▬ not yet — week-12 test pending |
| Are AI assistants sending more visitors? | Visits arriving from chatgpt.com per day | 0.31/day then 0.19/day vs 0.10/day before — 10 visits in 43 days vs 6 in the prior 60 | ▲ above baseline, both windows |
| Is brand demand on Google growing? | Brand-name clicks / day (locked query list) | 1.12 → 1.78 (+58%) across the upgrade; best sustained rate of the engagement | ▲ up — with caveats stated |
| Do you appear for treatment searches? | Non-brand appearances / day | Coverage of your actual services (pelvic floor, postpartum, dry needling) up ~3.4× — but at page-5 positions, so few clicks yet | ▲ warming — clicks wait on service pages |
| Is Bing's AI citing you? | Bing Webmaster AI citations | 0 this window (Bing directory listing still in progress) | ▬ quiet |
| Can we measure bookings? | Clicks from the site into Jane | Still zero recorded — the tracking gap diagnosed in July remains; intake-question fix pending | ✗ still broken |
| What else is driving traffic? | Everything in analytics | Google Ads stepped up ~54% in August — campaigns/keywords still unknown to us | ? need the ads detail |
| Lever | The pair | Baseline → current | Change |
|---|---|---|---|
| AI recommendations share of AI answers naming you |
with live search | 8.5% → 8.5% | ▬ flat — no gain claimed; week-12 test is the gate |
| from AI memory | 2.5% → 3.0% | ▬ inside the 1.0–3.6 noise band; all such mentions are Perplexity | |
| Brand search people looking for you by name |
clicks / day | 1.12 → 1.78 | ▲ +58% gate · +9% vs the June–July baseline (mid-July lull flatters the gate) |
| Google position | 2.35 → 2.39 | ▬ flat — the lift is clicks-per-appearance (6.4 → 8.9%), not rank | |
| Bookings clicks into Jane |
booking clicks / day | 0 → 0 | ✗ instrument broken — cross-domain tracking gap (diagnosed in July); intake-question fix pending |
Not yet — and we're not going to dress that up. Your profiles went live through mid-July and were bound to your identity on July 29; AI systems re-read and re-weight sources on their own clock, typically over one to three months. The pre-agreed test is the week-12 measurement (~late October) against the clean June 29 baseline. Until then: no claim.
The early tell we are watching: Perplexity — the platform that re-reads the web fastest — went from naming Joybody in 4 of its 50 answers to 7, and its citations of your own site nearly tripled. That's a handful of answers, so we call it a direction, not a result. It's the right direction.
| Run | Live search % | From memory % | Failures |
|---|---|---|---|
| May 5 | — | 3.6 | 6 |
| May 20 | — | 2.5 | 0 |
| Jun 15 | 7.0 | 2.0 | 0 |
| Jun 29 (pre-foundation anchor) | 8.0 | 1.0 | 0 |
| Jul 14 | 8.5 | 2.5 | 0 |
| Aug 25 | 8.5 | 3.0 | 0 |
"From memory" has bounced between 1.0 and 3.6 all engagement — we treat that whole band as noise. Every from-memory mention so far comes from Perplexity; ChatGPT, Gemini and Claude still don't know Joybody without searching.
Yes, against the baseline — 10 visits in this 43-day window vs 6 in the entire prior period. Before the citation work, AI assistants sent about one visitor every ten days (0.10/day). In the two windows since: 0.31/day, then 0.19/day. Both are at least double the old rate. The second window is lower than the first — with five visits per window that's normal scatter, and we show you both numbers rather than the flattering one.
This remains the portfolio's standout: healthcare customers click through to vet a provider before trusting an AI's recommendation. Which of these 16 visitors booked? We still can't see that — the booking-tracking gap below is the reason the intake question matters.
Brand demand hit its best sustained rate yet — 1.78 clicks/day, the fourth straight rise. The +58% jump across the upgrade is real but partly flattered by a quiet stretch in mid-July (you can see the dip in the fourth bar); against a broader summer baseline the lift is closer to +9%. What we stand behind is the staircase: every window of this engagement has beaten the last.
Three honesty notes. One: the profile go-live dates in mid-July are approximate (we're chasing exact dates), so "before vs after the upgrade" blurs two overlapping treatments. Two: August seasonality can't be ruled out on a first-year property. Three: your Google Ads stepped up ~54% in August — if any of that bids on the name "Joybody," the organic numbers above are understated. We still need the campaign detail.
| Window | Brand clicks/day | Brand appearances/day | Click rate | Position |
|---|---|---|---|---|
| Before (Apr 9 – May 6) | 1.29 | 15.9 | — | 2.4 |
| Corrective era (May 8 – Jun 29) | 1.43 | 14.8 | — | 2.1 |
| Post-foundation (Jul 1–12) | 1.67 | 21.1 | — | 2.6 |
| Pre-upgrade (Jul 13–28) | 1.12 | 17.5 | 6.4% | 2.35 |
| Post-upgrade (Jul 29 – Aug 24) | 1.78 | 19.9 | 8.9% | 2.39 |
The lift is people clicking more often when you appear (click rate 6.4% → 8.9%), not a ranking change. Total site clicks dipped 4.6 → 4.0/day across the same gate — that's entirely because the pre-upgrade window had an unusual hot streak on treatment searches; against the June–July norm, treatment-search clicks in August are actually up ~12%.
A leading indicator worth naming precisely: Google is now showing Joybody for the services we encoded in June — pelvic floor, postpartum, women's health, dry needling — about 3.4× more often than before the upgrade. These appearances sit around position 50 (page 5), so they produce almost no clicks yet. Appearances are coverage, not customers; the service-area pages on the build list are what turn this coverage into clicks.
| When | What shipped | What responded | Caveats | Evidence |
|---|---|---|---|---|
| May 7 | Corrective fixes — but the schema deploy silently failed (no AI-readable treatment existed) | Nothing, for 8 weeks — brand clicks drifted 1.29 → 1.43/day, AI answers flat | None — that's the point | Gated (control) |
| Jun 30 | Foundation (schema v3) genuinely live, verified in raw code + on-site fixes | Brand appearances +43% (14.8 → 21.1/day) and clicks +29% within 12 days | Jul 4–6 Facebook/newsletter days excluded — result holds | Gated |
| ~Jul 13–28 | Directory & social profiles go live progressively (dates approximate — being confirmed) | AI-assistant visits step to 0.31/day, then 0.19 (vs 0.10 baseline); Perplexity citing your site 3× more | Approximate dates · 5 visits per window · overlaps the Jul 29 upgrade | Directional |
| Jul 29 | Upgrade: map pin fixed + 6 profiles bound to your identity (verified) | Brand clicks +58% across the gate (+9% vs the June–July baseline); treatment-search coverage 3.4× | Mid-July lull flatters the gate · profile ramp overlaps · ads up ~54%, keywords unknown · August is August | Gated, with caveats |
| Jul 13–29 → ? | The same citation + binding work → AI recommendations (the headline lever) | 8.5% → 8.5% at day 27 — consistent with the 1–3 month lag AI systems take to re-weight sources | Pre-agreed test: week-12 measurement (~late Oct) vs the clean Jun 29 baseline | Pending — no claim |
| Metric | Start of engagement | July report | This report | The honest read |
|---|---|---|---|---|
| AI answers naming you (live search) | 8.0% | 8.5% | 8.5% | Flat two cycles running — the week-12 test in October is where this is judged |
| AI answers from memory | 1.0% | 2.5% | 3.0% | Bounces in a 1.0–3.6% band all engagement — we treat it as noise |
| Brand-search clicks / day | 1.43 (1.29 pre) | 1.67 | 1.78 | Up every single window of the engagement |
| Brand appearances / day | 14.8 | 21.1 | 19.9 | Down 6% vs July's figure — July's was a 12-day burst; this is 27 days sustained, still +34% vs the start |
| Brand Google position | 2.07 | 2.61 | 2.39 | July's growth-dilution partly digested; the click lift comes from click rate, not rank |
| Visitors from ChatGPT (running total) | 2 | 6 | 16 | The compounding line — 10 of the 16 arrived in this cycle's 43 days |
| Treatment-search appearances / day | 36.5 | 37.5 | 46.9 | +25% vs July — new coverage entering at page-5 positions; clicks wait on service pages |
| Services page appearances / day | 14.3 | 18.6 | 18.2 | Plateaued at ~18 — the next step change needs the service-area pages, not more waiting |
| Measurable booking clicks into Jane | 0 | 0 | 0 | 0 → 0 → 0 — the tracking gap, unfixed two cycles; the intake question ends this row |
| Bing AI citations (in window) | 4 lifetime | 1 | 0 | Bing's AI is quiet — its directory listing is still in progress |
| Identity readable in raw code | absent (failed deploy) | v3 live, verified | v4 live, verified | Re-checked every cycle — Wix can silently regress |
We can say: the citation network is built and bound to your identity in code, verified the way AI systems actually read it. Brand-search clicks are at their best sustained rate of the engagement — the fourth straight rise. AI assistants are sending visitors at double the pre-deploy rate or better in both windows since the profiles went live. And Google has started showing you for the treatments you actually offer.
We won't say yet: that AI recommends Joybody more (8.5% → 8.5% is flat, and we report it flat); that the +58% brand jump is proof (a quiet pre-window flatters it, the profile dates are approximate, and August is August); or anything about bookings — until the intake question ships, that number honestly doesn't exist. The claim gate is the week-12 measurement in late October. If the citation work is doing its job, that's where it shows.