Skip to main content

Is age just a number? The art of the algorithm

(Image credit: Image source: Shutterstock/Maxx-Studio)

The abundance of product and consumer data gives retailers the opportunity to transform their businesses, and the retail industry. Rich connected data provides opportunities to create true one-to-one personalised experiences. It’s clear that specific consumer attributes, such as age, and geography, have an impact on personal preferences. In turn, generating relevant experiences based on these insights about individual shopper behaviours has become a growing area of interest across the retail industry.

Providing one-to-one recommendations using basic, generalised shopper information isn’t true personalisation. True personalisation builds upon this basic information, taking the individual’s preference into account and offers recommendations based on those indicated preferences. Segmenting like-shoppers and offering them all the same group of recommendations is not personalisation, but generalisation. Until each shopper’s preference drives the recommendations they receive, true one-to-one personalisation can’t be achieved.

Take for example the question of age, does a consumer’s age act as the primary driver of personal preferences, or is it one of several factors that influences preference instead? Here, Rhonda Textor, Head of Data Science at True Fit discusses whether age does in fact drive a consumer’s personal preferences.

In the age of the consumer, nearly all decisions made by retailers are controlled by the shopper’s wants and needs. As more brands and retailers bolster their e-commerce experience, there is a higher demand to replicate the traditional brick-and-mortar (B&M) shopping experience. The consumer benefits of B&M shopping, such as the ability to try on clothes and shoes and style with other pieces, are difficult to replicate digitally. Consumers demand individualised digital experiences and push retailers to help them find recommendations based on their unique style, size, and fit preferences. In doing so, retailers can expect increased revenue, AOV, and consumer satisfaction, which leads to greater consumer loyalty.

A key topic of interest in generating true one-to-one personalised experiences for the consumer, and yet a controversial one, is age. A common industry approach is for designers, retailers and recommendation engines to segment people into averages and make assumptions about what a consumer wants to wear, based on their age.

But are certain apparel “attributes” that are traditionally geared towards older consumers (for example, longer hems for women and a less-fitted silhouette for men) reflective of older consumers preferences? Or does age merely influence these preferences, while other factors actually drive them?

Product & consumer data

Using a variety of different consumer demographic data retailers can offer accurate product recommendations that are based on the individual’s unique preferences. Consumer attributes can include age, body mass index (BMI), and geography, as well as previous purchase behaviours. The combination of this different data is when the recommendation becomes truly personal and relevant to the individual shopper.

No purchase history

When a consumer shops without previous purchase history data, retailers can predict which styles might align with their preferences, based on the consumer’s age, height, weight, and size. Once purchase history data from the consumer is available, the recommendations generated become specific to the individual.

Recommendations become more accurate when consumers make multiple purchases because a retailer is able to analyse their shopping behaviours. The more transactions an algorithm can analyse, the better recommendations they can serve, that are unique to the individual shopper.

Without any previous purchase history, consumers in the older demographic tend to opt for tops with more sleeve coverage, and the younger consumers purchase more sleeveless options, (based on aggregated data from consumers within the same demographic). 

However, in order to provide the best style, fit and size recommendations to consumers, you must consider a consumer’s brand affinities, purchase history and fit preferences. Taking all of this different data into account ensures that the individual shopper receives recommendations that are relevant and truly personalised.

While there is a pronounced preference for sleeveless tops or dresses among the smaller and younger shoppers, by incorporating purchase history data into the equation, many shoppers in the top quarter of size and age distributions also choose sleeveless styles, suggesting that age does not drive a consumer’s preference, but rather only influences it.  

With purchase history

By analysing consumer behaviours with purchase history data, retailers can better understand the shopper’s preferences. If a shopper’s age is the only factor being considered, then they may receive recommendations that are skewed to include items with more or less coverage. The extra layer of data, like purchase history, filters in more items that match the shopper’s preferences and are likely to be of interest to that shopper, rather than just assumptions made on their age alone.

There are several instances where older consumers may show an interest for shorter sleeve options or younger consumers start to purchase tops with more coverage. When a consumer’s purchase history shows a preference for sleeveless options, it is likely she will purchase more sleeveless options in the future. Algorithms learn to prioritise these preferences, expressed by the individual consumer, rather than defaulting to recommendations based solely on age.  With this data, retailers can target marketing, advertisements and recommendations towards the consumers who consistently indicate their preferences, for example, shorter or longer sleeve tops.

When recommendations are provided without the added data layer around preference, consumers may feel underserved or that they have been falsely recommended products that do not fulfil their wants and needs, causing a negative experience and weakened loyalty. The added data layer of purchase history ensures that every shopper receives an individually curated recommendation from the retailer.


When a consumer considers purchasing a product, specifically in the apparel and footwear industry, there are a variety of factors that can affect whether the consumer ultimately makes a purchase. The style, fit, and price of a product are three main considerations that consumers face when deciding which items best match their preferences.

Research shows that while a shopper’s age indicates which products they will likely be interested in, age alone cannot drive truly personal shopping experiences. Age is only a piece of the larger puzzle.

Individual style preferences are nuanced and require more than age and size to model accurately.  Data about a consumer’s age is important because it can help retailers gain an overview of the styles consumers among different demographics are more likely to love and keep.

Even with this small amount of volunteered information from shoppers, retailers can already provide relevant, personalised shopping recommendations. For shoppers who lack a sales history with online retailers (due to lack of online purchases or privacy concerns), providing a limited amount of personal data still results in relevant content.  Frequent shoppers who have a sales history can have a more personalised experience because machine learning algorithms can capture a more refined relationship with clothing features.

Retailers should look to use an algorithm that is able to curate product recommendations for consumers, with and without previous purchase history, allows them to power the best style recommendations to their consumers, based on demographic and product data. Age helps retailers guide their recommendations but cannot replace the value in generating recommendations based on each consumer’s expressed preferences.

To find out more about how a consumer's age influences his or her fit and style preferences download The Art and the Algorithm: A Consumer Behaviours Report here.

Rhonda Texter, head of data science, True Fit

Rhonda Textor has been the head of data science at True Fit, a platform dedicated to helping shoppers find clothes and shoes they love and keep, since 2015. She leads the company’s data science team and is passionate about modelling fit and style elements of both shoppers and garments to recommend products to shoppers that they love and that fit and flatter.