Ripple StrategiesRipple
First-party research · October 16, 2026

Our First AI Search Visibility Baseline

24 fixed prompts, two intended repeats, and the missing coverage made visible.

On October 16, 2026, Ripple Strategies tested its own visibility in anonymous desktop AI-product sessions using 24 fixed BC buyer prompts, with two planned runs per prompt on Gemini, Perplexity and ChatGPT. We captured 57 valid answers across 144 planned slots. Ripple was absent from the 42 unbranded Gemini answers and eight measured unbranded Perplexity answers. This is our first baseline, not a client result, a search ranking or evidence of change over time.

Start a project

At a glance

01

48 Gemini answers

All 48 planned answers had complete source captures: 42 unbranded answers did not mention or cite Ripple; all six branded answers did both.

02

9 Perplexity answers

Eight unbranded answers did not mention Ripple. One branded answer mentioned it and linked to its site. Citation capture was incomplete for all nine.

03

87 unmeasured slots

Perplexity had four account gates and 35 unrun slots. ChatGPT's 48 planned runs were unrun after access challenges. None count as measured zero visibility.

Ripple Strategies AI visibility baseline — October 16 Pacific / October 17 UTC, 2026
Product and cohortValid answers / plannedRipple mentionsComplete source capturesAnswers with observed Ripple links
Gemini · unbranded42 / 420420
Gemini · branded6 / 6666
Perplexity · unbranded8 / 42000 observed; absence unconfirmed
Perplexity · branded1 / 610At least 1
ChatGPT · unbranded0 / 42UnmeasuredUnmeasuredUnmeasured
ChatGPT · branded0 / 6UnmeasuredUnmeasuredUnmeasured

What questions did we ask?

The frozen set contains eight marketing, eight website and eight software prompts. Each family has seven unbranded discovery, comparison or buying questions and one question explicitly naming Ripple Strategies and ripplestrategies.ca. That creates 21 unbranded and three branded prompts, each with two intended repeats per product.

Examples include which SEO agencies serve the Fraser Valley, which companies build small-business websites in Chilliwack, and which BC software agencies publicly explain ownership and support. The exact, unchanged wording and prompt IDs are in the downloadable prompt CSV. These questions are a defined BC buyer panel; they do not cover every possible query or represent a randomized sample of buyers.

How were answers and citations measured?

Captures used logged-out desktop product interfaces and fresh conversations. They were taken October 16 Pacific, which falls on October 17 UTC in the timestamped records. The prompts name BC markets, but physical searcher location was not established. Browser state was shared within the anonymous session, and provider-side personalization was not controlled.

Gemini displayed Flash-Lite. Its search mode was not established. Perplexity used its default Search interface; its underlying model was not displayed. We used actual product answers, not API completions. Gemini source chips were opened and their displayed links captured; Perplexity source expansion remained incomplete. Earlier exploratory Gemini captures without complete source expansion were excluded from this formal cohort.

A mention means Ripple Strategies or its domain appears in the answer, excluding the user's prompt. A target citation means an observed source URL has the ripplestrategies.ca hostname, normalizing www; another similarly named domain does not count. A brand mention, a recommendation and a factual answer are different observations. Only complete source captures can establish citation absence.

What did the first baseline show?

Gemini returned all 48 planned answers. Ripple appeared in none of the 42 unbranded answers and was mentioned and cited in all six branded answers. Recognition when the question names the business does not demonstrate that the business will be recommended when a buyer asks for providers.

Perplexity returned nine valid answers: eight unbranded with no Ripple mention and one branded with a mention and at least one Ripple link. Four submitted requests reached account gates, and the remaining 35 slots were not run. All nine valid answers had incomplete citation captures, so an unobserved Ripple link cannot be reported as proven citation absence. ChatGPT's 48 planned runs remained unrun after access challenges.

The six summary rows and all 144 slots are available below. Unmeasured slots carry their status without invented mention or citation values. The source-link CSV contains the public URLs actually observed, including incomplete Perplexity captures; a source's presence is not an endorsement or an audit of that provider.

Did the branded answers describe Ripple accurately?

Not consistently. One branded Gemini marketing answer described retired organic social-media management as current, repeated unsupported aggregate performance claims, and overstated the absence of named team members and published ad-management pricing. We compared those statements with Ripple's current full FAQ and About page. We are not republishing the unsupported figures as business results.

At the time of comparison, the About page named Dallen Oakley and Adair Chavez and described the two-person studio. The FAQ distinguished paid advertising from organic social-media management and published ad-management fee information; SEO was individually quoted. A citation to a business page does not by itself establish the answer's accuracy. This observation does not identify whether caching, source selection or another mechanism caused the mismatch.

What does this mean for SEO, AIO and GEO work?

The practical starting point is useful, accessible service information supported by inspectable work and consistent business facts. This baseline gives us a defined panel to repeat after improving those pages. It does not show that a particular file, schema change, directory placement or content format causes an AI citation.

A follow-up should retain the prompts and repetitions, record product and session context again, disclose access failures and compare equivalent cohorts. More complete coverage is required before drawing conclusions about Perplexity, and ChatGPT has no measured baseline here. Qualified inquiries and attributable referrals should be measured separately from mentions and citations.

What can this study not establish?

It cannot establish Google organic position, local-pack rank, Google AI Overview or AI Mode visibility, market share, a universal engine ranking, or a change in revenue. Gemini is a different product from Google's search AI features. Two repeats can also share underlying retrieval and response patterns; they are not independent proof of a stable recommendation rate.

This is research on our own brand, conducted by the business being measured. It is not an independent evaluation or a client outcome. The downloadable tables disclose the prompts, dates, context, coverage, observed links and measurement rules; they contain no private Search Console data, customer records or long copied AI answers.

Frequently asked questions

Was Ripple recommended in any unbranded answers?

Ripple was not mentioned in the 42 unbranded Gemini answers or eight measured unbranded Perplexity answers in this first panel. Unrun or blocked requests are unmeasured. A brand mention is also distinct from a positive recommendation.

Does this study show improvement from AI optimization?

No. It is an initial baseline on Ripple's own brand, with no comparable earlier panel or untreated control. It reports observations, not a measured gain or a causal effect from an optimization.

Why are some citation results marked incomplete?

Perplexity's nine valid responses were captured, but all source expansions were incomplete. A positive Ripple link was observed in one branded answer; absence cannot be established for the others. Only Gemini's 48 complete source captures support citation absence or presence across that full panel.

Talk through your project.

Book a call