Pet Food Observatory: how AI influences pet parents’ choices

An industry survey to listen to pet parents and read the LLM “black box” to design better experiences

Pet food is a sector where needs are recurring, but choices are rarely simple. It’s a purchase that’s repeated over time, but often seen as a highly responsible decision: ingredients, origin, sensitivity, ethics, sustainability, expert advice, and a comparison of alternatives all factor into the evaluation.

In this context, we launched the TSW Pet Food Observatory: a place of continuous listening, created to collect and interpret the real experiences of pet parents and transform them into insights and operational directions to improve content, touchpoints, and the quality of the experience throughout the customer journey—including emerging channels, such as conversations with LLMs (Large Language Models).

According to the Assalco–Zoomark 2025 Report (2024 data), the dog and cat food market in Italy will reach €3.1 billion in 2024 (+3.7% vs 2023), and the pet population is estimated at approximately 65 million animals. This sector is particularly interesting for us also because pet parents are a highly informed and engaged target: they compare sources and tools along the journey and build trust over time, purchase after purchase.

Why a pet food research Observatory (and why now)

When we talk about experience, we’re talking about relationships: between those who produce and those who choose, between what a brand communicates and what a person understands, expects, and decides. In pet food, this relationship is even more delicate, because the choice affects the well-being of a family member.

Today, a new point of contact is added to all this: conversations with AI. More and more often, people don’t just “search,” they engage in dialogue. They ask for advice, comparisons, alternatives, and reassurance. And they do so within conversational environments that generate responses, build narratives, and—in fact—can influence expectations and brand shortlists.

Hence the question guiding the Observatory: how does the pet parent’s choice experience change when the LLM comes into play? And what does this mean, concretely, for brands that want to be relevant and credible?

The project: four lenses on experience

We structured the Observatory pilot into four complementary phases, designed to take an initial snapshot of the sector and operationalize listening.

1) Qualitative phase: listening and interviews with pet parents

We start with people. Through qualitative listening and interviews, we reconstruct needs, drivers of choice, frictions, languages, and contexts: what really happens before, during, and after the purchase (when repurchase comes into play).

2) BARTT: measuring associations between brands and values ​​(beyond what’s declared)

We complement the qualitative phase with the BARTT – Brand Association Response Time Test, to measure explicit and implicit associations between brands and values ​​(e.g., quality, price, ethics, naturalness, sustainability). This is a useful lens because it helps us understand not only what is declared, but also how immediate and “rooted” certain associations are.

3) Analysis of LLM conversations: understanding how the industry is spoken

Here we come to the newest point: we look at how LLMs currently speak about the pet food industry. In other words: what happens when a pet parent asks for a “tailored” recommendation in a conversational environment? Which brands are mentioned? In what situations? With what motivations and trade-offs?

4) Analysis of communication on more traditional touchpoints

Finally, we analyze how the industry communicates on more traditional touchpoints, such as social media, but also the spaces and content where authority is typically built and choice is supported. We compare language, formats, and messages with what emerges from listening and conversations on LLMs, to identify misalignments and opportunities for improvement along the journey.

Opening the LLM “black box”: new conversational intents

One of the pilot’s distinctive contributions is the analysis of LLMs with an approach that doesn’t simply replicate SEO. We categorized new intents typical of conversational search, in which the user doesn’t type a query but engages in dialogue, providing details, correcting, asking for reassurance, and comparing options.

This reading of the intents allows us to shed light on three key aspects:

• What the pet parent really asks for, contextually and situationally;
• How the LLM constructs the industry narrative (values, categories, trade-offs);
• Where brands are strong or weak: when they are mentioned, in what way, and with what associations.

We conducted the analysis using a proprietary method and tools, simulating and observing repeated conversations to identify patterns and recurrences.

The pilot context: the buyer persona “Laura”

To make the dynamics observable and compare responses consistently, we worked on a simulated buyer persona for the pilot:

Laura, 38, new owner of an English Setter. She is an administrative employee, married, and the mother of a 7-year-old child. She lives in a provincial town in Italy (e.g., Emilia-Romagna), has an average household income (about €2,500–3,000 net per month), and is a first-time dog owner.

This is a useful profile because it well represents a widespread condition: a person who wants to make the right choice, who seeks guidance and discussion, and who alternates sources and tools along the journey.

What emerges: the brands most cited by LLMs and in which contexts

The observed conversations reveal citation and positioning patterns that help us understand how the industry narrative “works” within LLMs.

Here is a summary of the most cited brands and the main context in which they were mentioned:

• Farmina (N&D / Vet Life) – very high frequency: often cited as a recurring reference for life-stage kibble, veterinary diets (e.g., renal, struvite, gastrointestinal), and functional lines (e.g., quinoa, pumpkin).
• Monge – very high frequency: direct and consistent alternative, often paired with Farmina for maintenance, single-protein, hypoallergenic, and veterinary lines.
• Edgard & Cooper – high frequency: associated with sustainability and snacks (treats, dental), as well as wet food and “fresh meat” or grain-free options.
• Almo Nature (HFC) – high frequency: frequently cited for wet food, natural supplements (e.g., salmon oil), light biscuits, and holistic/alternative diets; Often also as a “topping” or for dogs with picky palates.
• Royal Canin – medium frequency: recommended primarily for specific medical needs (Veterinary Diet lines such as urinary, renal, hypoallergenic) or for sporting/active dogs, often as a “technical” alternative.
• Yarrah Bio – low frequency: cited when the request focuses on organic, vegetarian, or highly sustainable products.
• Others (Hill’s, Terra Canis, etc.) – very low frequency: sporadic mentions related to niche needs (e.g., sensitive stomachs, wet puppy food).

Beyond the mentions: examples of insights that make data “useable”

The most interesting part is not just “who gets mentioned,” but how they’re framed and for what needs.

Some brief examples (among those that emerged in the pilot) that demonstrate the type of insights we can extract:

• Almo Nature tends to emerge as a “versatile” brand when it comes to requests related to naturalness/transparency, palatability and hydration (e.g., wet/pouch), light snacks, and supplements. Conversely, it’s less frequently cited when the demand shifts to technical-veterinary needs or performance.
• Royal Canin is often positioned as a clinical or para-veterinary reference: it’s strongly cited for needs such as renal, urinary, gastrointestinal, and hypoallergenic. However, it’s less associated with “clean label” or sustainability requests, where the LLM orients toward alternatives perceived as more “natural” or “ethical.”
• In comparison conversations, the narrative tends to polarize between approaches perceived as more holistic/natural and others perceived as more scientific/clinical. This is a significant pattern because it influences expectations about efficacy, trust, availability, and even the “justification” of the price.

How we use these insights: from research to strategy, to experience

The Observatory wasn’t created to produce a report for its own sake. Our goal is to “deploy” research to help brands:

• improve the quality of information (clarity, completeness, verifiability, consistency);
• reduce friction and uncertainty along the journey (before and after purchase);
• design a coherent ecosystem between content, touchpoints, and relationships over time;
• understand and manage their presence in emerging touchpoints, where initial convictions and shortlists of choices are formed today.

In short: transform listening into operational, measurable, and prioritized insights.

A pilot to demonstrate a replicable approach

The work presented is a demo/pilot project: it does not aim to exhaust all case studies in the sector, but rather to demonstrate how a structured analysis can transform listening into actionable insights to improve content, touchpoints, and the quality of the experience.