Report
Answer consistency
Sessions
Syd 7 · Syd/Melb 10 Jul
Market
Australia
Assistants
ChatGPT · Perplexity · Gemini · Claude
Report · answer consistency

We asked ChatGPT the same question four times. The answers never matched.

The question was simple: “Best physiotherapist in Sydney – who do you recommend?” We asked it four times in one sitting, and ChatGPT answered confidently every time.

Of the 23 businesses it named, 13 appeared once and never again. Only one was named in all four answers, and ChatGPT wrote its name four different ways doing it.

The recommendation survived. The name barely did.

The stability band, in one breath

Ask an AI the same question four times and the answers shift. The band is that range, and a change only counts as movement once it clears it.The method in full →

Not even consistency is consistent.

ChatGPT went 0.23 to 0.56 with nothing changed but the date. Gemini agreed 0.58 on “best in Sydney” and 0.30 on the same question three days later. Consistency moves with the question, the market and the day.

fig. 1 · how much each AI agrees with itself
0.00.51.0.56.50.31.34.59.58.30.44.89.90.92agreed just .23ChatGPT · Sydney · 7 JulChatGPTPerplexityGeminiClaude
Syd · 7 JulSyd · 10 JulMelb · 10 Julthe story reading

ChatGPT agreed just .23 – Sydney, 7 July.

ChatGPT
.23
.56
.50
Perplexity
.31
.34
.59
Gemini
.58
.30
.44
Claude
.89
.90
.92
Sydney · 7 JulSydney · 10 JulMelbourne · 10 Jul

0 = never the same answer · 1 = identical every time

The grey is the noise; the lume is the one reading the data will stand behind.

agreement with itself · 0 (never) to 1 (identical) · three sessions · stability pilot v1 · Sydney 7 Jul, Sydney repeat and Melbourne 10 Jul 2026
download · png · svg
1the setup

What we did

On 7 July 2026 we asked two buyer questions: the Sydney physio question above, and a near-Bondi variant. Four assistants, four times each, one sitting. 32 answers, asked the way a customer would ask.

We count businesses, not strings. The methodology covers what counts as one, and the dataset carries every name exactly as each assistant wrote it.

Ask the same question four times and the shortlist barely holds. The grid below is what each assistant returned, ask by ask.

fig. 2 · best in Sydney
assistantnames per answerbusinesses namednamed onceagreementnamed in all four
ChatGPT11 · 15 · 4 · 92313·57%0.23SportsFit Health and Rehab (Five Dock)
Perplexity6 · 6 · 5 · 7136·46%0.31Andrew Noyes – Exact Physiotherapy
Gemini9 · 10 · 8 · 7134·31%0.58Align Health Collective · Evoker Premium Physiotherapy Services · PhysiCo City · Sydney Physio Clinic
Claude6 · 5 · 6 · 560·0%0.89Evoker Premium Physiotherapy · Fixio · PhysiCo City Physiotherapy (Sydney CBD) · Sydney Physio Clinic (Macquarie Street) · Sydney Physio Solutions
agreement · 0 (never) to 1 (identical) · stability pilot v1 · Sydney 7 Jul 2026 · entity extraction v2.1 · name variants merged, both counts in the dataset

The same pattern held on the second question, a suburb-level ask where the lists were shorter but no steadier.

fig. 3 · near Bondi
assistantnames per answerbusinesses namednamed onceagreementnamed in all four
ChatGPT8 · 4 · 6 · 7126·50%0.42ISO Physiotherapy · JW Physical Health
Perplexity3 · 3 · 3 · 352·40%0.53JW Physical Health
Gemini4 · 7 · 4 · 483·38%0.51Bondi Platinum Physio · Physio Fitness (Bondi Junction)
Claude4 · 3 · 1 · 352·40%0.45The Running Room Physiotherapy
agreement · 0 (never) to 1 (identical) · stability pilot v1 · Sydney 7 Jul 2026 · entity extraction v2.1 · name variants merged, both counts in the dataset
2the count

Named once, never again

ChatGPT answered the same question with 11 names, then 15, then 4, then 9. You can be recommendation #3 in one answer and absent from the next.

23
businesses named across four answers
13
named once and never again
1
named in all four answers
3the survivors

Who survived

One business came through all four of ChatGPT’s answers: SportsFit Health and Rehab, written four different ways. The variants ran from the full name with its suburb down to a bare “SportsFit”, and all four are in the dataset.

Across assistants, Sydney Physio Clinic was named in every answer by two of them, and again by Claude in the Sydney repeat three days later. Evoker came through all four asks for three separate assistant-and-day pairings.

This is not a ranking. Appearing or vanishing is the assistants’ behaviour, not a judgement of any practice.

One limit, plainly. Identity is keyed on a business website domain where the assistant links one, and on the name everywhere else. Sydney’s CBD has near-identical physio names, so name-matched counts carry that caveat.

The assistants read largely the same pages and still named different businesses. Being cited and being named are two different numbers.

fig. 4 · agreement · businesses vs sources
assistantagreement on businesses namedagreement on sources cited
ChatGPT0.230.36
Perplexity0.310.61
Claude0.891.00
Gemini0.58n/a – unresolvable redirects
agreement · 0 (never) to 1 (identical) · stability pilot v1 · Sydney 7 Jul 2026 · source agreement by website domain
4the implication

The sources hold still. The names don’t.

If one question returns 23 different recommendations in four asks, a single screenshot of an AI answer is not evidence of anything. An AI ranking published without a range is noise dressed as signal.

This is why the methodology attaches a band and a date to every reading, and why The Southlume Index publishes rates with bands rather than bare ranks.

The recommendation survived. The name barely did.

5caveats

Open questions

This is a pilot, and we say so. Two questions, one city, four asks per assistant, repeated in Melbourne and again in Sydney three days later.

Melbourne ran three days after Sydney, so that comparison changed the day as well as the city. We re-ran Sydney on the same day as Melbourne to separate the two.

The answer refused to be tidy, and we publish it anyway. An assistant’s consistency isn’t stable with itself from one day to the next. A snapshot is a snapshot, even if you take four of them.

on the recordquestion set stability-pilot v1 · physiotherapists · Sydney 7 Jul 2026 · Sydney repeat and Melbourne 10 Jul 2026 · 2 questions × 4 assistants × 4 asks, one sitting · entity extraction v2.1 · name variants merged, both counts published in the dataset · 22 Jul 2026: counts restated under entity extraction v2 · 25 Jul 2026: restated under v2.1 – a first-word filter was dropping real businesses whose names begin with their suburb; every prior figure is preserved in the dataset.

The Southlume Index is measuring these categories now.

notify me · edition 1 →see the methodology →

Prompt checks are a sample; referral analytics are the ground truth of actual AI-driven visibility. A zero here is ‘not found on these prompts’, never ‘invisible’.