Shopper marketing blog

Consumer-facing brands should leverage bespoke consumer feedback data to train or fine-tune AI / ML models to get ahead. These expert models can help them gain valuable insights into consumer preferences, behaviours, and sentiments, enabling them to optimise product descriptions, packaging design and advertising copy.

One application we have developed for such AI models is to assess and classify product reviews with helpful tags. Using consumer feedback as the basis, we have created proprietary measures of product reviews, such as how influential an individual review is and how strongly it reviews the product's taste or effectiveness.

By analysing these new measures, brand owners can identify their performance and patterns in consumer reviews, generate new insights, compare themselves to competition and even highlight individual reviews for a specific purpose, such as "suitable for advertising". 

Using bespoke shopper data to train AI models represents a significant opportunity for consumer-facing brands to enhance their understanding of consumers and markets.

AI consumer Measures use case: Review Performance Assessment

As part of reporting and analysing consumer product reviews, we have gone beyond the usual bland sentiment analysis. Instead, we asked ourselves, "Which measures of reviews would we find useful?" The first was "How influential is this review?" So, we set about creating it.

We put thousands of carefully selected product reviews before our online shopper panel. We asked them to rate them based on "How likely is this review to influence you to buy this product?" (they knew the product category but not the exact brand). We gained multiple ratings for each review, tens of thousands in all. 

After QC, we trained AI LLM models on their answers. After several rounds of tweaking, we delivered models that agreed strongly with shoppers. We then could rate all our reviews (we collect product reviews from sites worldwide) against this question and many more. The AI models then tagged each review with a score or label as appropriate. Analysing this new data provides some fascinating insights.

Brands can identify which products perform well and which may require further attention. This information can guide brands' decision-making, helping them allocate resources effectively and prioritise product improvements.

In a previous blog post, we used our "Influence" measure of review to illustrate the variation in individual impact of reviews across the moisturiser category. The table below shows the reviews in UK supermarkets from March 2024.

From a quick look, Olay is doing very well, with plenty of reviews and a good "Average of Review Score". However, when the Influence Score is added to the mix, it is clear that the Olay reviews are much less influential than their competitors, and in fact, L'Oreal is doing much better.

Creating measures like Influence Score can give brands fresh insights into performance and areas of opportunity. Why not let us know what measures you would like to create?

Utilising Bespoke AI Measures for Product Insights

Another way of utilising bespoke measures for product insights is the ability to uncover consumer preferences. From such data, brands can identify the specific features, attributes, and benefits that resonate with consumers. This information can be used to develop products that better meet the needs and desires of the target audience, ultimately driving customer satisfaction and loyalty.

We have created universal "Taste" and "Effectiveness" measures in reviews. We can use our consumer-based trained AI models to tag reviews based on how much they talk about Taste (for food and drink items) or Effectiveness (for Health and Beauty or Household items). 

Bespoke measures can also highlight areas for improvement in existing products. This is like creating your own panel of shoppers answering exactly the questions you want to ask. Client-based analysts can then identify common complaints, concerns, or consumer suggestions. This information can guide brands in making data-driven decisions on product enhancements and optimisations, leading to improved customer satisfaction and loyalty.

Competitor Analysis with new AI review measures

A common use of our AI-tagged review data is identifying areas of competitive differentiation. Companies can identify market gaps and innovation opportunities by comparing and contrasting data from different brands. This information can guide brands in developing unique selling propositions and positioning strategies.

For example, as a Health and Beauty brand, you could look at your sub-ranges in terms of Effectiveness head to head with your competition highlights where you need to focus on R&D.

Brands can identify the key features, attributes, or benefits that resonate with consumers and drive sales and we can help them create their own AI measures. This new knowledge can inform product development and marketing strategies, helping brands better meet the needs and desires of their target audience.

Selecting Reviews for Advertising with AI

It has become increasingly common to see genuine consumer reviews used in brand marketing campaigns. The reviews are used as social proof or "reason to believe" to support the campaign's main message. Here, Inch's Cider used reviews to underline its message of attracting new cider drinkers to their brand.

We have a "Suitable for Advertising" measure, making bringing these reviews to the fore easier than ever. Add to that measures tagged with Influence or Taste, and then Shopper, Marketing and Analyst teams can sift and sort reviews to find the content they need.

What can you do?

Leveraging bespoke shopper data to train AI models represents a significant opportunity for consumer-facing brands to enhance their understanding of consumers and markets. By using our measures or creating their own brands, agencies can identify their performance and patterns in consumer reviews, generate new insights, compare themselves to the competition, and even highlight individual reviews for specific purposes such as advertising.

Using new AI measures for product insights enables brands to uncover consumer preferences for specific features, attributes, and benefits, ultimately driving customer satisfaction and loyalty. The possibilities are endless, and we encourage brands to consider creating their own measures to gain fresh insights into performance and areas of opportunity.

If you are a brand or marketing agency, why not make it easier to deliver more influential campaigns by surfacing exactly the reviews you need? 

CheckoutSmart can create bespoke measures, tag your data, analyse your data, provide tagged data and even create new reviews if required.

What are you waiting for? Come join the data revolution.