Faculty Forecast: AI trends for retail & consumer
In January, we published our first edition of the Faculty AI Forecast for 2021 – a broad look at the AI trends likely to dominate business and society over the coming year.
In it, we highlighted three forces that will shape AI adoption in the next 12 months:
- Public trust and data ethics remain critical.
- Organisations think in terms of AI systems, not AI models.
- Organisations make getting quality data a priority.
Now, we’re diving deeper into how these three trends will be reflected in some of Faculty’s specialist industries.
In this blog, Faculty’s Director of Retail & Consumer rounds up the trends he expects to see across Retail, eCommerce, Consumer Goods, and Travel and Hospitality.
Consumers and companies demand digital marketing that delivers on its promises
Digital marketing is suffering from a trust crisis.
On the one hand, consumers are increasingly uncertain that brands and tech companies can be trusted with their data – and that the brands they do trust with their data will offer any meaningful reward in return.
On the other hand, companies are becoming sceptical that online platforms can deliver on their promises of granular customer insight and staggering ROI. It’s taken a while for the hype bubble to burst, but many brands are beginning to realise that the ‘next generation’ capabilities promised still often mean brands waste money advertising to people who would have bought anyway. On top of this, the death of the third-party cookie – including their removal from Chrome by 2022 – and changes made by Apple in iOS and Safari are increasingly signalling a death-knell for the days of unfettered online tracking and ad targeting.
In the face of this change, brands will begin – and, in fact, already are – demanding tools that help them realise the promises of digital for both their brand and their customers.
In practice, that means many will continue to move marketing spend away from platforms that suck in huge amounts of customer data, keep most of it for their own ‘walled gardens’, and use the rest to create personalised campaigns of dubious effectiveness.
We also expect to see a move away from ‘out-of-the-box’ solutions that offer an immediate lift in conversion at the expense of long-term growth. Brands are recognising that such tools can cause reputational damage by invasively over-targeting certain customers whilst leaving other customers under-served.
Instead, we’ll see more companies looking for long-term, sustainable incrementality – solutions that gain value over time by growing in-house first party data assets and use them to increase performance, personalisation and the insights generated over time.
Brands adapt their demand forecasting to handle data-related uncertainty
When – or if – the UK economy reopens completely on 21 June, it’s unlikely that we’ll snap immediately back to pre-COVID business-as-usual. Instead, it’s highly likely that we’ll enter a period characterised by a new, different form of data-related uncertainty when it comes to planning for consumer demand.
Uncertainty is difficult for businesses, whether it’s associated with an economic upturn or a downturn. Carlsberg are anticipating a “jazz age boom” this summer as restrictions ease. This is great news for the economy at large and for companies like Carlsberg in particular – but how should they quantify uncertainty and plan properly for it, so they can avoid lost sales opportunities, premium shipping fees and excess storage costs?
Just as there was no useful, granular data on how your consumers’ behaviour changes during a pandemic, there’s no data on how a country’s recovery from a pandemic affects consumer behaviour either. As a result, it’s unlikely that sales data from 2019 – or even from 2020 – will be able to accurately predict demand in the latter half of 2021.
As a result, many brands will look to tools that go beyond past sales data to incorporate a wider range of sources, like high street footfall or consumer search activity. Indicators like these – that can be updated almost in real time and more accurately reflect the wider economic context in the here-and-now – will be useful additional early indicators of demand that, in turn, will help brands respond to changes as they unfold. .
While integrating these additional data sources presents a grand opportunity, there are also a host of new challenges. Teams in retail & consumer are well versed in predicting future sales from past sales data, but understanding how promotions, events, or weather patterns should be integrated into demand models is relatively new terrain.
Brands recognise that building better customer experience isn’t really about having good AI – or even a good data science team
A 2020 Algorithmia survey found that, among companies actively engaging in machine learning, 55% had not yet deployed a machine learning model.
Just as hype for online marketing is beginning to sink into cynicism, many consumer-focused organisations are beginning to realise that hiring a few data scientists and throwing data at them isn’t the fastest road to AI adoption.
However, many organisations have fallen into taking a ‘model-first’ approach to data science in order to justify this investment. This approach centres around building the models that most obviously fit the available data, rather than those that are most valuable to the business and that integrate well with existing business processes.
The result is that many brands are still struggling to get their AI initiatives past the proof-of-concept stage. Those that do get their models into production often find that this takes much longer than expected, that the resulting live model / application is ignored. or that maintaining performance requires significantly more effort than anticipated.
Over the next year, we expect this frustration to spur a new approach; one that sees AI adoption as a business-wide transformation, not a software installation. As a result, many will begin to think more holistically about their AI implementations and building a more multidisciplinary approach to development.
For many, this will mean finding partners that can support areas of AI implementation where businesses lack experience or resources themselves. These partners might help organisations to develop in-house data science teams, build strategies, or simply take on day-to-day operations and maintenance on an ongoing basis.
Discover more AI trends for 2021
If you’d like to discuss how AI can transform your business, our Retail & Consumer team would love to hear from you – get in touch here to kick off a conversation.