Case study · Workflow automation + AI integration

How a Singapore physiotherapy clinic became visible in AI search.

A multi-therapist practice with a strong offline reputation was near-invisible when patients asked AI assistants for care — and slipping on Google. We built a managed SEO + GEO engine that changed both in weeks.

Client: A Singapore physiotherapy clinic (anonymized)SEO + GEO managed serviceAI-visibility engineStructured dataRank tracking

Outcomes at a glance

+83%AI-assistant referral visitsin about four weeks

Page 1a priority treatment keywordreached position ~6

100/100Structured-data healthfrom a near-zero baseline

18Prompts trackedacross Claude + Gemini

Metrics from a live NexSprint engagement. Client anonymized at their request; figures rounded for readability.

The problem

Where things stood.

The clinic is a well-regarded, multi-therapist physiotherapy practice. Offline, its reputation was strong. Online, it barely existed where new patients were increasingly starting their search: inside AI assistants.

Ask ChatGPT, Perplexity, or Gemini for a physiotherapist in Singapore and the clinic almost never came up. Competitors did. On classic Google search it was drifting down the rankings for the terms that mattered most.

Two root causes compounded the problem. The site carried no structured data, so search engines and AI models had little machine-readable signal about what the clinic did or where it operated. And there was no way to measure AI-search visibility at all — you cannot improve a number nobody is tracking.

What NexSprint built

Human-led, agent-built, iterating weekly.

Rather than a one-off audit, NexSprint stood up a productized SEO + GEO managed service: a client portal that runs continuously, surfaces the highest-ROI fixes first, and measures the one thing legacy SEO tools ignore — whether AI assistants actually mention you.

Automated site scans

Weekly and monthly crawls score the site across technical health, content, schema, and performance — flagging regressions and ranking the fixes that move the needle first.

AI-visibility engine

A tracker runs 18 real buying-intent prompts across Claude and Gemini on a schedule, records whether the clinic is cited, and charts its share of voice against competitors over time.

Structured-data engine

A schema layer emits clean, valid structured data — organization, medical practice, services, location — giving both Google and AI models the machine-readable signal that was missing.

Keyword rank tracking

Core physiotherapy service terms are tracked on a schedule so movement on Google is visible week to week, not guessed at.

Client scorecard dashboard

A single portal view rolls every signal into one scorecard the clinic owner can read in a minute — no SEO jargon required.

Weekly iteration loop

Human-led and agent-built. Each week the agent team ships the next batch of fixes, re-scans, and reports — so the service compounds rather than stalling after launch.

The outcome

What changed.

Within about four weeks, referral visits arriving from AI assistants were up roughly 83% — real patients following an AI recommendation to the clinic's site.

On Google, the clinic reached page one at around position six for a priority treatment keyword, one of its most valuable terms. Structured-data health went from effectively nothing to a perfect 100/100.

The per-assistant picture tells the clearest story. Visibility share on ChatGPT roughly doubled from 5% to 11%; on Perplexity it went from 0% to 25%; on Gemini from 0% to 20%. A practice that AI assistants used to skip is now part of the answer.

Because it is a live managed service rather than a project, the numbers keep moving. Each weekly cycle re-scans, ships the next fixes, and reports — so visibility compounds instead of decaying.

AI-visibility share · before → after

ChatGPT5% 11%

Perplexity0% 25%

Gemini0% 20%

Tech stack

Next.js portalCloudflare WorkersCloudflare D1ClaudeGeminiStructured data / JSON-LD

Timeline

Live managed service with continuous weekly iteration. Headline movement landed inside the first month; the engine keeps running.

Your turn

Want the same, measured against your own baseline?

We start with an audit, agree on the KPI, and ship a pilot. If we miss the number we agreed on, you don’t pay for it.

Start an audit See more work