Topview of office desk with documents and laptop. Businesswoman working on paperwork.

Case Study: AI-Powered Guest Insight at Fox Harb’r Resort

AI slashed research costs by over $20K and delivered actionable insights in just ten business days.

Blog

by Peter Coish

Latest from the Blog

Reports of Journalism’s Death are Overstated

We’re finally realizing that the “free news” era was an unsustainable detour.

The Consumer Isn’t a Moron, She’s Your Bot

Within a decade, shopping won’t revolve around people tapping on ads. That’s bad news for ad agencies like us.

From Broadband to Battlefield

Fibre optics is an example of transformative tech that makes lives better. But not in Ukraine.

Perplexed About Perplexity?

It’s time to pay attention to Perplexity and its potential as an ad platform.

Ad Tax Blows Up Trade Negotiations

DST is yet another cost borne by Canadians and a sore point with the Orange Man.

Our cherished client Fox Harb’r Resort wanted hard numbers on guest satisfaction and clear guidance on how to sharpen its digital targeting ahead of the 2025 season. In the past we would have hired a third party research shop to handle this project, paid somewhere between $20,000 and $30,000, and waited a couple of months for the results.

This year we kept everything in-house and let AI do the heavy lifting statistically-speaking.

Our Tech Stack

  • Mailchimp – distributed a survey with over 200 data points to 9,000 past guests and piped the 500+ responses to a database

  • OpenAI Deep Research + o3 model – data cleanup, statistical computation, and narrative drafting

  • Looker Studio – interactive dashboards for ad hoc exploration

  • Google Docs – final research decks

The only cash outlay was for subs to Mailchimp and OpenAI’s ChatGPT Pro (it’s mind-blowing); everything else was labour.

Objectives

  • Quantify guest satisfaction through Net Promoter Score (NPS)

  • Build an actionable segmentation for precision digital advertising

  • Identify attitudinal and service gaps among golfers, non-golfers, and golfers who also play at competitor courses

Survey Design & Launch

Our survey mixed Likert scales and a handful of open-text prompts. Conditional logic in Mailchimp ensured golfers and non-golfers saw context-specific questions, keeping average completion time under ten minutes.

Data Intake & Cleaning

Responses were fed to Deep Research, where the o3 model normalized data, standardized free-form text, and flagged outliers.  We went for a coffee.

Crosstabs Generation

Using o3-generated SQL based on our natural language questions, we compared:

  • Golfers vs non-golfers

  • Golfers who have played only Fox Harb’r vs those who also play competitor courses

The model surfaced any statistically significant variations in NPS, reasons to book, and propensity to book return visits. And a whole lot more. We reviewed and green-lit every auto-generated table.  More on that below.

We also uploaded the data to Looker Studio dashboards to let us do additional ad hoc queries on variables like  visit frequency, state/province of origin, or golf behaviour.

Narrative drafting

o3 then wrote first-pass commentary. Us sentient humans at Kuration then edited (substantially) for accuracy, sense and voice.

This is critical. Hallucinations are inevitable. Skip this step and you risk presenting something ridiculous to your client.

Cluster Analysis

Here where AI really starts to blow our minds. Remember cluster analysis from that third year stats course? No? We don’t blame you.

Well, it’s no longer the exclusive domain of statisticians.  Now stats laypeople (like us) can run them.  We used Deep Research to run such an analysis (for the propeller-heads following at home this was a k-means clustering that tested for 2-7 clusters).

The analysis identified four distinct guest segments. Each cluster was assigned a persona with along with a description of lifestyle, demos, and media habits. We even used Dall-e to generate pics of each persona!

Deliverables

  • Two detailed results analysis decks including NPS trends, golfer/non-golfer insights, etc. with recommendations for management action
  • Since each deck was more than 5,000 words, we also included executive summaries (AI-generated, natch)
  • A deck with a one-page description of each of the four segments

Results & Cost Advantage

  • Our client had decision-ready insight just 10 business days after the survey closed

  • Hard costs were less than $800 (Mailchimp and ChatGPT Pro subs)

  • We saved our client tens of thousands of dollars versus a traditional third-party engagement for an NPS survey and cluster analysis

Kuration’s ad ops team has already loaded the four-segment taxonomy into Google Ads and Meta to sharpen the performance of campaigns on these platforms.

Takeaway

AI has moved survey research from a massive capital expense to an operating habit. Instead of commissioning an expensive third-party study, Kuration can now spin up an end-to-end insight study at near-zero marginal cost. Deep Research handles text analytics, o3 handles reasoning, and we humans handle judgment. That means Fox Harb’r Resort pays for brains, not busywork – and gets actionable management info that goes well beyond a simple NPS score. Fast.