The rise of generative AI, social e-commerce and the streaming wars: Deloitte experts on the TMT trends to watch in 2023

  • After Deloitte’s recent Technology, Media & Telecoms (TMT) Predictions Breakfast in Belfast, Sync NI caught up with Paul Lee, the company’s Global Head of Research for TMT and Ben Stanton, TMT Insights Manager, for a rundown on some of the key tech trends they expect to gather pace in 2023.

    Paul, you talked last year about the likely introduction of adverts in streaming services and that’s now happening. Do you think this is now the norm?

    Paul: A year ago that’s what we expected to happen because people’s budgets were tightening a little bit, so we expected that if consumers were offered the option of watching a few ads and saving a few pounds, they would go for it. But what has happened in the interim is a significant hit to incomes caused by inflation, particularly with energy and food, so behaviours have probably shifted more quickly than we expected. One of the trends we’re seeing as a result is a high degree of sharing of memberships, people saving money by getting together with friends or family to share subscriptions. For some services that’s ok, but others have put in controls on doing that. This year one of the big challenges is to what extent will people accept not sharing their subscriptions. In the UK, 55% of households pay for a video subscription but the reach of the subscriptions is about 80%, so there’s quite a gap there.

    You made a related prediction to do with streaming services paying big money for the rights to major sporting events. Do you see this as the next battleground for subscribers?

    Paul: There is a play here about attracting additional users and there’s also a play about retention. Churn in Subscription Video On Demand (SVOD) is a big issue because it’s really expensive to add a user, but it’s really easy to unsubscribe. So having ‘must watch’ content like sports is a way of attracting and retaining. The challenge in that market is that delivering a stream of sports is hard technically, because you can’t compress. If there is a delay then for a big football, rugby or cricket match, the tolerance for that delay is much lower. With video-on-demand, you can have a delay of over a minute, which is not going to be acceptable for the big games. So how to control latency is a big challenge. The second thing is these services are trying to become profitable and this is a significant cost, so it’s a difficult balance to maintain. It could work out well, but it is a high risk, high stakes strategy.

    Ben, one of the other major trends you see evolving is social media e-commerce, with spending up significantly on social channels like TikTok and Instagram. Can you tell me more about that?

    Ben: The prediction is that social commerce is going to be a trillion-dollar industry this year. It’s being driven by young people and the trick is really in bringing together the point at which someone decides to make a purchase with the ability to make that purchase. If you decide to buy something but have to flick through two or three websites to get to a storefront, it adds friction. If you’re able to be inspired to buy something and then conduct a transaction in the same place, it removes the layers of friction. A lot of young people, who have had a social media account for their whole lives, are comfortable doing that. They’re very susceptible to this too because you have a group of very attractive people online doing the selling. The social e-commerce model really was pioneered in China and various versions of it are being exported worldwide. We see lots of places it is being adapted and lots of excitement, for example various players in Africa are interested in how their young citizens will buy things in future. The storefront is the social media platform and so a large amount of the commerce that happens on social media is impulse buying. In most cases that’s different to a transaction with a retailer where the customer has an idea of what they want to buy before going on the platform. But our forecast is certainly for social e-commerce to grow quickly.

    Paul, you weren’t convinced the metaverse was going to be the gamechanger for Virtual Reality (VR) last year. Is it any closer to coming into its own?

    Paul: With VR, for it to work, you have to have content. That’s the case with any device. But as of now, content creators are not flocking towards VR. If you look at the smartphone, there are about five billion units in use, and probably today 90% of them will be used and the median usage will be about three hours. If you’re a content creator with a game or app, you’ve got a target market of effectively four or five billion. If we look at the customer base of installed headsets, currently it’s about 15 million globally – equivalent to about five days of smartphone sales. When you have quite a small user base, it’s hard to attract content creators. If a major application of VR is games, the question then is what sort of gamers will you attract? You have casual gamers and then you have those who play for hours that need powerful machines to attach to it. What you have with VR would be a powerful processor attached to quite an imprecise controller. Will that work? I don’t think it will. If you don’t have regular usage you don’t get growth in the market and you don’t attract content creators.

    You had another prediction this year about the affordability of smartphones improving, so will that market keep growing?

    Paul: Smartphones are very good at lots of things today, and I also think they’re very good at lots of things for tomorrow. For example, having passports stored on smartphones. In some states in the United States, the driver licence, a primary form of ID, can reside on the phone. Car keys can reside on an app on the phone. The same utility for a fraction of the cost and I think there is a lot of utility for the smartphone still to be delivered. When it comes to price point, you can now get very low priced 5G phones but it’s also worth considering that a £1,000 smartphone in two years might be £500 and in five years’ time it may still be useful and £200.

    Ben: When we look at devices, every category that has shown significant growth over the last five years has been adjacent to the smart phone. Things like blue tooth earbuds, hundreds of millions of units sold, high margin products that make a lot of money. Smart watches too. Those orbit around the planet that is the smart phone.

    So, let’s turn to generative AI. What’s your take?

    Ben: One of the reasons I think generative AI has become such a viral sensation is that it creates things for people that are instantly shareable and it makes users feel special as they create content. Even if it is something as simple as writing a different bedtime story for your child every night with their name in it, people will share that on social media.

    Paul: The cut through for generative AI is really mainstream. There are some technologies like the metaverse that skew towards specific demographics, but with many new generative AI tools it can be accessible to the 11-year-old doing homework all the way to people in their 70s and 80s. The reason they cut through is because the application is obvious – doing things faster – and the user interfaces are often really easy. Any technology that thrives has to have good underlying technology and an accessible user interface.

    Ben: What I think we’ll see over the next year is more examples of high-quality data sets being used in a private context – companies that have their own data sets that want to build their own models, not for public consumption. It could be something as simple as creating a much better chat bot for customer service, or providing FAQs from a website that can be answered from 100 different angles and questions asked from 100 different angles, to give original answers and outputs, to increase customer satisfaction. One of the terms you’ll hear a lot more in the next year is prompt engineering.

    Paul: If you type in artificial via a search engine it guesses your next word might be intelligence. A lot of the new and emerging generative AI tools are just a more complex version of that. We have so much AI already in our lives, this is just a new version of it.

Share this story