Skip to main content
All CollectionsInsightsMethodology and other information
Mintel Leap methodology and coverage
Mintel Leap methodology and coverage

Find out more about the technology behind Mintel Leap and understand which content and data is covered by this AI solution.

Updated over 2 months ago

Mintel Leap is our unique AI-powered insights engine, built exclusively on Mintel's proprietary research and expert human analysis. Leap revolutionises your Mintel experience by using our content to answer the questions you need to drive your business forward.

Keep reading to learn all about the tech and data behind Leap.

💡 Tip: Learn more about how you can get the most out of your Mintel Leap experience by reading this article (for Leap subscribers only).

The content and data Leap uses for its responses

So far, Leap has ingested Mintel's analyst insights that you can find on clients.mintel.com, including reports and trends. That's over 150,000 pieces of content, exclusively from our own experts! In order to give you the most up-to-date information, Leap references insights from the last 30 months to date. Where an insight is published regularly, such as every year, Leap's responses always reference the latest version of the insight.

At the moment, Leap's responses do not include consumer data from our Databooks (where that data is not explicitly mentioned in a report), GNPD product launch data and data from other Mintel solutions, and custom content created for your organisation.

To summarise:

Ingested by Leap

Not (yet) ingested by Leap

Insights on clients.mintel.com, written by the Mintel and Comperemedia analysts

Custom content created for your organisation only

Consumer data from Databooks

GNPD product launch data

Data from other Mintel and Comperemedia solutions

We are continually working to improve the capabilities of our AI, and the quality of its responses. Check back regularly and see Leap's capacity grow!

How Leap works in detail

The large language model (LLM) we are using currently to power Leap is a version of ChatGPT.

Mintel Leap works by semantically searching for your query through all of the content that we have indexed. It then identifies the 10-15 closest blocks of Mintel content in semantic meaning to your question. The LLM will only use those blocks to answer your question – Leap does not reference any information outside of Mintel.

💡 Tip: In case Leap cannot find any Mintel content related to your query, you'll see a message to let you know Leap could not answer that particular question. Try asking it in a different way – you can find tips for improving your Leap queries here on our Help Center (for Leap subscribers only).

Leap generates new responses according to the prompts you give it. There is an element of variance in the way our LLM works, just like there is for all LLMs. This means that you can sometimes see small variations in the response, even if you ask the same question multiple times. If your query is more ambiguous, there can be bigger variance in what Leap decides to summarise. Even if the response varies for the same question, it will always be grounded in Mintel expertise.

In some cases, you will see Leap's response broken up into numbered segments. The segments are ordered by their semantic closeness to the question you asked. This is the same for the list of insights mentioned in the Sources. There is some probability-based variance programmed into all LLMs to enable them to be more creative in their answers, so you may see some small change in the order with similar questions.

As ChatGPT gets better at various tasks, we will benefit from its improved functionality due to the way we've built Leap. This way we are leveraging the best of Mintel and our ever growing content, and also the best of LLMs in the shape of ChatGPT.

📌 Note: We are not training any LLM on our content and data, nor are we providing your questions to the LLM. Any tracking that we do only happens so that we can monitor the quality of Leap's answers. That means Mintel Leap doesn't become naturally smarter – instead we are continuously improving the underlying search to ensure Leap understands your queries better, and plan on adding more content and data to make this solution even more powerful. Since we own all of the content that Leap uses to generate its answers, we feel we are in an ethical safe space. If there ever are any concerns about the ethics of ChatGTP, the way we've built Leap means we could replace our LLM with one from a different provider if needed.

🧭 Up next: If you're already a Mintel Leap client, read this article and learn how you can use it to boost your research on any topic you care about.

Did this answer your question?