Will Beauty Be The Proving Ground For Personalisation?

This article was originally published in The Interline’s AI Report 2025. To read other opinion pieces, exclusive editorials, and detailed profiles and interviews with key vendors, download the full AI Report 2025 completely free of charge and ungated.

---

Social media feeds lit up recently  with the announcement that OpenAI and Shopify had partnered to bring what they call a new frontier of personalised shopping to the masses. This news followed not only OpenAI CEO Sam Altman’s claim that 10% of the world currently uses their system, but also the revelation that Google searches fell in Safari for the first time ever last month, primarily due to end users now searching with generative AI instead of traditional search.

Together, these signals point to a potentially significant shift: consumer journeys are gradually moving away from the common experiences and narratives shaped by traditional search engines, and for the first time, toward hyper-personalised results, scoured from across the internet and delivered in conversational, relatable, and unique responses powered and generated by AI. Some of the foundations of this have been laid by algorithmic social feeds (nobody sees the same thing in their home feed as their friends, after all), but AI is now ratcheting up the personalisation another notch.

And on that basis, this coming together of OpenAI and Shopify represents more than just a strategic and well-considered collaboration; it signifies that deeper change to how we engage with the internet is coming, and for the first time since the dawn of ecommerce, the way we discover, interact and shop for products online is in an actual state of flux.

How different categories could deliver on the vision for hyper-personalisation

Outside of enterprise circles, a lot of people are surprised at just how quickly consumer AI adoption has scaled. On top of ChatGPT becoming the most popular consumer application in history, generative AI has entered the zeitgeist in a major way.  And for all the cultural questions it raises, consumer uptake of AI surfaces a longstanding home truth: that the so-called “endless aisle” of the internet is no longer what people actually want when they open a browser or an app. Instead, customers desire direct, curated and no-fuss results–needs that have gone unfulfilled in an era of information overload, volume, variety, and a firehose approach to aligning what brands make with what shoppers actually need, mediated through the arcane art of attention-based ad bidding.

And now that the masses have the tools to enact more agency as consumers, we’re already seeing evidence that personalisation is a common preference. A lot of people would simply prefer to have a chatbot mediate their interactions with the web, than to spend time sifting through Google results. That matters for the people making websites, obviously, but it matters equally – if not more – to the gigantic share of brands that sell through the web.

While there’s obviously a variety of opinion about the value of AI in general, within fashion, forecasters, strategists, and writers like myself are facing a deep question: we’ve long championed the need for hyper-personalisation as a remedy to product saturation, shouting about the benefits of recognising audiences as individuals with their own set of drivers, intentions and expectations… is AI now the right delivery mechanism for those essential ideas? And does that offset some of the reservations we have about it?

In the first instance, it’s fair to say that personalisation hasn’t really taken flight to date, without AI. And who could blame brands and retailers when, beyond tailored marketing, there are so many barriers to positioning such a granular focus on individual customers? When garments can be endlessly adjusted and refined to the particular needs of a consumer before being sent into a narrow, brittle funnel of sourcing and manufacturing, hyper-personalisation quickly becomes a wicked problem across workflows that the industry hasn’t managed to crack.

For the moment, AI isn’t proposing to solve the upstream side of the equation for garments, but, with the convergence of AI sophistication and consumer openness to adoption, it does feel like hyper-personalisation is becoming feasible at the consumer end.

And if there’s an industry that is better set up to then go and execute on the creation of personalised products, it’s more likely to be beauty than fashion.

Why? Because not only does the beauty industry have a reputation for disruption, but it’s just far simpler to draw boundary lines for the limitations of a beauty product. Where a jacket, for example, can vary in size, colour, use cases, and fabrication, a moisturiser, can only vary its formulation as it strives to achieve one purpose—healthy and hydrated skin.

Take makeup brand Dcypher, as a real-life example, the company uses an AI-powered skin tone measurement system to create made-to-measure foundations matched to each customer’s specifications. This is manageable because it works within limited parameters such as finish, coverage, skin type, and with input from an AI skin-tone scanner powered by Perfect Corp. And we are already beginning to see more such examples across the industry as we move away from generic to a customisable ecosystem of products, experiences, and relationships delivered from brands straight to us as individuals.

Consumer appetite for this kind of personalisation in beauty is also already apparent: a recent Mintel report found that 62% of US beauty and personal care buyers are interested in hyper-personalised products, and 28% are willing to pay extra for them. And since factors like social media have placed the beauty industry at the centre of both popular and niche culture, and at the forefront of digital natives’ minds, it’s little wonder that personalisation at the product level–rather than just the routine–is resonating with so many.

A new era of beauty consumer

The absence of almost complete subjectiveness found in the fashion industry provides beauty with a much more informed and discerning customer when it comes to personalisation. Just explore TikTok, Instagram or YouTube, you’ll inevitably come across young consumers discussing the intricacies of a skincare routine tailored to their problem areas, or posting comments that lament the inclusion of a new ingredient that’s been added to their favourite concealer. While individual outcomes vary, and claims are open to criticism, the beauty community at large focuses on objective attributes and results.

Unlike in fashion, where so much relies on taste, trends, and fast consumption, a high proportion of these young, digitally literate consumers understand facts, with a science-backed knowledge of product compositions, sustainability credentials, and specific ingredients to look out for based on their own desired outcomes.

As a result, any brand that incorporates personalisation into their product offering must provide results that are sophisticated, sensitive, and grounded in genuine science and data, not surface-level marketing that is all claims with no backing. The average beauty customer knows what they want, and it is therefore essential that brands walk the line between expertise and an openness to learn if they are to build relationships with consumers that respect their knowledge, values, and individual goals.

Sephora does this exceptionally well. With a growing youth audience known as “Sephora Kids”, the retailer is currently attracting a much younger demographic who are starting their personal care journeys far earlier than the generations before them. While the retailer has always been ahead of the curve, launching its AI-powered Virtual Artist back in 2016, it was their observance and agility when they noticed customer tendencies to undertake more detailed product searches that led them to serve AI-generated landing pages based on the user’s search terms and profiles.

Not only can Sephoranow meet customers where they are–responding to their desires rather than dictating need or relying on assumptions–but they can also help to reduce the time, effort, and energy needed to discover the right product without customers even realising they’ve been assisted.

The future of phygital appearance

More overt AI usage within a beauty context is already a staple of everyday use for the average 16–to 25-year-old as they straddle digital and physical worlds. From personalised avatars and professional headshots, to beauty try-on filters, our phones are offering us new lenses from which to view and manipulate ourselves in a variety of contexts.

FFFACE.ME, for example, is a creative studio focused on enabling brands to integrate technology into their marketing and communications strategies. Having already worked with Fenty Beauty, Maybelline, and Prada Beauty, they have a rich portfolio of AI-powered social media filters, 3D models, and virtual try-on experiences that allow brands to connect their products with individual customers in real-time, indicating a consumer desire for brands to embrace the fun, exaggerated, and interactive elements of beauty.

But, as consumers continue to desire personalisation across physical and digital channels, we are seeing the emergence of much more pragmatic tools drawn from this technology. With consumer expectations for consistent customer service, paired with product accuracy, beauty brands will increasingly need to draw a distinct line between the avant-garde and the realistic, with a knowledge of how and when to show up for their audiences, either as an immersive experience or an informative seller. And accurate try-on experiences and simulations are essential to the latter.

“We’re not simulating what users want to see, we’re simulating what’s realistically achievable based on their skin baseline and scientific evidence. That commitment to realism is core to how we support brands and build trust with consumers.” CEO & Co-Founder of HAUT.AI– an AI-powered skin analysis solution–Anastasia Georgievskaya tells me.

Her words are a stark contrast to how AI has been typically deployed until now, working to ‘improve’, ‘refine’, or drastically alter our appearances, playing with the boundaries of what resembles our original selves. Yet, the growing importance of accuracy is not to be underestimated, as brands look to use these methods to convince customers of a product’s suitability as they increasingly connect and sell through digital channels.

So as phygital solutions continue to operate with a sense of duality to not only mask our appearance but also uncover our true selves, beauty brands will now need to balance fun and engaging interactions by also going skin-deep to offer skin analysis and simulations that understand the realities of our faces, moving beyond the typical use of filters that simply sit products on top of us but instead seek to forecast their impact and results as we use them.

Beauty as a bridge between self-care and science

I would therefore argue that the most significant development in beauty’s ongoing relationship with AI might be its ability to take us from the current landscape of surface-level filters or the generation of uncanny valley-generated selfies to physical and tangible changes to appearance in the real world.

And how beauty and skincare impact our physical appearance has become an essential consideration for the modern consumer. Against the backdrop of accessibility to healthcare and information becoming more democratised, recent research suggests that 82% of Americans consider wellness to be a top priority in their daily routines, and in the same survey, 74% of respondents prioritise self-care and wellness in their beauty rituals.

Lately, I have also seen many beauty and skincare brands declaring a shift to putting “science back into beauty”. While I was unaware it ever left, I believe this points towards a change in how we talk about and consider beauty products as tools enveloped into our pursuit of such wellness, health and improvement. Roughly half of UK and US consumers reported clinical effectiveness as a top purchasing factor, while only about 20 percent reported the same for natural or clean ingredients.

This correlation between health and beauty is part of the reason why we have an influx of brands using biometric data to enable holistic approaches to how we manage and optimise our bodies. Within biometrics is a rich pool of opportunity to draw from the existing processes and technologies utilised in the health and wellness sectors to understand the more contextual aspects of our lives and appearances.

When questions like:

• How much sleep did you get last night?
• What’s the air quality like where you live?
• What was your sugar intake yesterday?
• And are you currently experiencing any hormonal fluctuations?

can be readily answered through the convergence of data collected via everyday devices and wearables like Apple Watches and Oura Rings (which are themselves rolled up into broad health apps and platforms, some of which offer API integrations to allow other applications and services read / write access to them), brands can begin to learn how these factors impact each of us as individuals with greater speed and accuracy to support product recommendations and even custom formulations.

La Roche-Posay have leaned into this with their Spotscan+ Coach app, developed in partnership with the mental health app Calm. It is a 3-month blemish-prone skin programme that pairs skin analysis with a personalised daily plan to help users consider the broader variables that affect their skin, such as stress. Then there is L’Oreal’s  ‘Perso’ device, launched back in 2020, which uses geodata to assess environmental factors such as weather, temperature, pollen, UV index, and humidity, to analyse and then dispense personalised skincare formulas.

Brands, therefore, should look to leverage the opportunity that connectivity between this ecosystem of datapoints presents and be conscious to contextualise their consumers with broad datasets if they are to truly make impactful formulations and accurate recommendations that help to guarantee product success and meet the modern beauty customer’s expectations.

Personalisation in-store and at home

Although rigid in their need for visible results, beauty consumers today have become increasingly flexible in their openness to shop across channels.

It wasn’t that long ago that even the thought of buying a foundation without trying it on first seemed inconceivable, but today, it is an everyday practice thanks in part to the maturity of virtual try-ons, skin analysis, in addition to globalised markets and expanded product ranges for diverse skin complexions, types, and textures.

In fact, until relatively recently, personalisation has borne the hallmark of a luxury in-store experience. And it is clear to see why, with individualised communications, one-to-one consultations, and bespoke-made formulations, these types of experiences used to take a considerable amount of time, money, and resources to deliver.

Launched back in 2016, Lancôme’s Le Teint Particulier, was initially designed as an in-store service that scanned a customer’s skin in three different locations to precisely analyse the type and shade of foundation they require, to then build a custom foundation formula that matches a customer’s skintone. Yet, the brand has now shifted focus to scaling various digital solutions that enable customers to engage with their products from anywhere. From matching existing Lancome foundation shades using a database of 22,000 skin tones to AI-driven skin analysis that recommends products drawing from 40,000 skin pictures the brand is leveraging the ease of smartphones and channel agnosticism to offer personalisation to broader audiences.

While it is not clear if the brand is still investing with the same tenacity in in-store integrations of technology, it is clear to see why targeting more scalable and accessible personalisation might be the short-term priority. With lower barriers to entry and a keen focus on match-making existing stock with customers more closely, these digital solutions feel like more manageable first steps into hyper-personalisation guided by AI.

But this is not to say that more real-world solutions aren’t feasible. In contrast, digitally native emerging brands and startups are starting to see the value that meeting customers in the real world, both in-store and at-home can provide.

BoldHue, for example, uses a scanning method similar to Lancôme’s, with the added benefit of analysing your skin shade from the comfort of your own home. After finding your ‘perfect match’, the device allows users to mix their own foundation with refillable cartridges, making a once-stressful shopping process more convenient, repeatable, and truly personal. Then there’s Mink’s 3D beauty printer, which offers the ability to print custom makeup on demand, pushing the boundaries of individual expression and product customisation even further. It seems now, however, that the brand has leaned into the experiential and novelty aspects of the solution, focusing on partnerships, events, and activations instead of individual at-home use.

This shift toward hyper-personalisation with at-home devices is not limited to makeup; tools like Remington’s StyleAdapt Technology are built to learn from users’ unique attributes, adapting their performance with each use to deliver increasingly tailored results.

Adoption, however, may remain the key challenge to such at-home solutions. Often requiring long-term brand loyalty and confidence in product performance for consumers to buy into these devices. It is therefore vital that brands don’t forget to pair the design of more seamless user experiences with products that genuinely deliver. It will not be enough for customers to purely access any shade of eyeshadow from a printer; it must also meet a broader spectrum of expectations, such as longevity, impact on skin and finish, if individuals are to invest in tools that create these products at home.

Many consumers may also still value the intervention and expertise of a beauty consultant throughout the customisation process, making in-store personalisations and one-to-one interactions a service that brands should continue to consider. If anything, current customer sentiments alongside the challenges of personalisation at scale highlight that the future of beauty tech will not be dictated solely at-home or in-store but will lie in accuracy, ease, and perhaps most importantly, trust in product performance over novel experiences.

A new landscape for beauty brands, large and small

Decision fatigue is pervasive across all consumer sectors, but beauty’s reputation as a “recession-proof” industry means the market has become oversaturated with choice, and the influx of eager startups has left existing brands needing to navigate a competitive landscape of low consumer retention.

With new players left and right, offering science-backed solutions, cutting-edge technology, and more inclusive approaches, it is fundamental that heritage brands diversify if they are to remain front of mind with consumers. While it is common knowledge amongst consumers that legacy beauty brands’ products are generally ‘good’–they’ve done the research and made the investment–there is a shifting mindset towards customers asking, ‘Is it good for me?’

Georgievskaya explains that this was the thinking that led her to develop HAUT.AI, “Back when I was working as a researcher in a lab, I kept noticing that people were using well-formulated, clinically tested skincare products, but they were still unhappy with the results. And it wasn’t because the products didn’t work. It was because they weren’t the right products for their skin.”

And this means whole suites of AI-powered solutions are becoming indispensable to large and legacy beauty brands. From skin analysis tools, virtual try-ons, to AI-generated visualisers, these options offer easy wins for brands looking to apply the principles of personalisation to existing product offerings where individual customisation is too heavy an investment.

But this is just the customer end of the funnel, and it seems that heritage brands are already working to embed AI deeper into their organisations to inform how they provide these experiences more inconspicuously “…when AI is part of their strategy, (legacy brands) are not limiting AI to consumer-facing experiences, they’re embedding it deeply within R&D, clinical testing, and product innovation.” Georgievskaya claims. And this is a promising assessment for the future of product customisation, which will require dramatic operational, cultural and creative shifts in the existing processes of large companies. Such investment in strategic applications for AI signals a pragmatic approach to how the technology streamlines internal barriers to customisation.

On the other side of the spectrum, emerging brands have the agility to build hyper-personalisation into their product range from the ground up, working with customers to fine-tune offerings at a much more granular level, growing alongside these needs rather than retrofitting solutions to them. Although this competition between depth of integration and speed of application is stiff, it does look as if the future of beauty could become a rich ecosystem of brands that learn from each other as they pilot and deploy new innovations from rapid-fire testing to carefully considered R&D.

This will be promising new ground for any tech partners with the foresight to tailor their offering to the broad spectrum of B2B client needs. What will be essential throughout, however, is that brands of all shapes and sizes stay true to their own vision and audiences by implementing AI in ways that genuinely meet the needs of their audiences and find the right collaborators to get them there.

Can hyper-personalised beauty truly be for everyone?

To answer the overarching question of this piece, it should first be acknowledged that true personalisation must go hand-in-hand with inclusivity, diversity, and accessibility if it is to effectively and accurately speak to each of us as the individuals we are.

We should then consider that both beauty and technology have historically been notorious for their lack of inclusive practices– there are countless examples to prove my point.

So within this context, I’m curious to see how these two components come together to achieve what they couldn’t do for so long by themselves.

Because of this, it is key that diversity and ethics stay front of mind throughout the development of new AI-powered solutions in the beauty industry. “One common issue in the industry is that many algorithms are trained on narrow datasets, which can lead to biased results.” Georgievskaya tells me before going on to explain HAUT.AI’s commitment to rectifying such issues by building their products upon diverse comprehensive datasets, using “over 3 million facial images across diverse ethnicities, age groups, and even rare dermatological conditions”. And they’re not alone in this consideration of diversity, startup Hue, for example, provides brands and retailers with user-generated content consisting of honest reviews to demonstrate how products work across different skin types, tones and concerns, enabling them to broaden their data-sets moving forward. Then take Carra Labs for instance, they use AI to deliver personalised recommendations for textured and curly hair care, softening pain points felt by often overlooked customers.

But Georgievskaya reminds us that there are still limitations, “ even with the best data, AI isn’t magic – it’s a tool. And people need to understand how it works…It’s not enough for a model to make a recommendation; it should also explain why it made that recommendation. This is exactly what our new Deep CARE recommendation system is designed to do. It doesn’t just suggest a product, it shows the user how the AI arrived at that choice. This explainability builds trust.”

And I am inclined to agree with the potential power of “explainability” within AI. Beyond building trust, it is essential that brands have discourse with their audiences, to build feedback loops that offer customers the opportunity to understand why and how certain conclusions have been drawn and respond in real-time. Even if brands are already pulling from diverse datasets, customer input must be folded in order to scrutinise models and improve solutions if they are to truly align with customer desires.

The future of AI-powered beauty tech

The myriad of emerging beauty-tech solutions makes me optimistic about the industry. Not only are smaller brands and startups driving fair and innovative solutions, but they also co-sign their audience’s desire to be themselves, and as a result, legacy brands will also have to lean into catering to their audiences as individuals and all the diversity that ties into that if they are to compete now and in the future.

However, at this point in time, as with any new innovation, widespread adoption is often marked by an immediate inaccessibility for the average person. So while I do not doubt that the cost of more experimental granular hyper-personalisation in the guise of custom formulations and at home devices will likely pass onto the consumer in the form of higher prices in the short-term, in the longer-term, AI should allow science-backed and data-driven personalised beauty persevere helping it to become more time-efficient, cost-effective, and scalable.

For now, there is a sliding scale of options, making different levels of personalisation accessible for almost everyone. Whilst fragmented for now, I look forward to the day when these come together more seamlessly to understand the more nuanced aspects of our appearances and can deliver an experience, journey, and results designed for me and only me.