Friday, April 12, 2024
HomeFinanceMatalan using AI to tell you how your top will look—and to...

Matalan using AI to tell you how your top will look—and to convince you to buy



“Add a cool and casual option to your wardrobe with this black T-shirt. Crafted from soft and cozy cotton, it features a classic crewneck and short sleeves for a comfortable fit. Finished with a NYC print to the front, it’s perfect for teaming with your favorite jeans or shorts.” 

So reads the description of an NYC-branded top on the U.K. retailer Matalan’s website. 

Thousands of these brief descriptions are dotted across the internet, intended to give customers a better idea of what they’re buying and implicitly giving way to a brand’s message. 

It’s also simultaneously the first step in getting shoppers to the website and the last one in convincing them to make a purchase. 

Those descriptions have typically been the domain of an ambitious fashionista picking up odd jobs early in their staff career at a retailer, or a budding copywriting intern doing the grunt work before they move up to tougher tasks.

Now though, it’s becoming the work of a bot.

Matalan’s AI pivot

In what Matalan described as a U.K. first, the company says it was able to “quadruple” productivity by using generative AI to help in-house copywriters create product descriptions on its website, with the help of London-based tech group Kin + Carta.

As Kin + Carta’s retail and travel portfolio director Matthe Hildon puts it, Matalan wasn’t satisfied with the speed at which they were able to get products out to market as they waited for descriptions to be completed. 

Then came the frustration with products not being found online and customers needing to be convinced to make a purchase once they got through to the site.

Matalan also struggled with a high level of attrition among its in-house team of three copywriters, who found the task of writing thousands of descriptions so mundane they were often driven to quit, Hildon says.

“We talked about the power of AI and whether it could be used to take an image, take a description, very basic data about the product, and whether we just wanted to do a proof of concept to see if we could try and create some very powerful product descriptions,” says Hildon.

Kin + Carta’s bot can create longer product descriptions at a faster rate than human writers, and doesn’t succumb to terminal boredom while doing so. 

The model developed its parametric memory from internet sources, in addition to customer data to understand the Matalan tone of voice, as well as its typical shopper, according to Kin + Carta’s chief technology officer in data and AI Karl Hampson.

The main question Hampson sought to answer was: “What can a model generate from its own memory, from his own understanding, and what can it generate from the customer data that you bring into the equation?” 

While the base model was “pretty good,” the secret sauce, as Hampson describes it, came from incorporating that customer data and honing prompts to fit a Matalan voice.

“The general trend here is one of making it easier for us to find and buy products in retailers catalogs, and that’s only a good thing for the retail industry,” Hampson says.

Matalan’s in-house copy team will continue to oversee descriptions, the retailer said, fact-checking things like the material of products and editing for clarity or Matalan’s tone of voice. 

Eventually, Hildon thinks the bot could improve productivity 20-fold by dramatically increasing the output of the copy team.

Another benefit Matalan is hoping to see from the use of AI is in its products’ ranking on Google search.

Google ranks links to websites based on their usefulness to searchers, elevating or demoting companies based on the feedback of customers. Hildon and Hampson hope AI will be a game changer for answering shoppers’ queries.

“Karl and I are describing this year of 2024 as the year where we’ll see a complete change in search behavior,” Hildon says.  

Finally, Kin + Carta’s developers hope their product description model will do the final job of converting visitors into buyers, as retailers chase the holy grail of a 100% purchase conversion rate.

How copywriters are adapting

Tasks like those Matalan is automating have typically been the domain of copywriters, who like other content creators are likely to have been keeping an eye on the progress of AI since ChatGPT exploded onto the tech scene last year.

Colm Hebblethwaite, a senior writer and AI lead at U.K. copywriting agency Stratton Craig, was one of those copywriters cognisant of the threats posed, and opportunities to be created from AI.

He set up an internal task force last year to assess how Straton Craig might be able to use AI to improve productivity and adapt to changing demands from clients thanks to the tech.

Hebblethwaite described product description jobs, like those being automated by Matalan, as low-quality, high-quantity tasks that are ideal for automation. 

If Matalan’s venture into automated product descriptions is a success, agencies like Stratton Craig can probably expect fewer requests from retailers for that high-quantity work.

Stratton Craig had previously pivoted away from this content to focus on higher-quality work like ESG reports and white papers. 

These can’t be replaced by automation yet, but Hebblethwaite has found other ways to utilize the technology.

The copywriting agency uses a platform called Jasper, alongside Microsoft Copilot, which helps it perform more backend activities, allowing them more time for creativity.

Hebblethwaite, for example, will ask his copilot specific questions about his interviews using the transcripts.  

The copywriter says his productivity on certain tasks has clearly been improved by the technology, and he’s optimistic that will remain the case.   

“We’re going to have to adapt, but the essential rules of the game for us, high-quality content and great strategic thinking, haven’t changed.”



Source link

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments

Translate »