Sri Sharma is the co-founder of Increasingly, a company harnessing the power of AI to drive revenue growth via ecommerce cross-sales. The Commerce Collective speaks to him about how the technology works and its impact on the future of retail
Sri Sharma is unwavering when asked what inspired the formation of Increasingly.
“Our inspiration, quite frankly, was Amazon,” he says. “If you look at a product page on Amazon, you’ll see the product and directly below it, you’ll see a bundle; below that, you’ll see product recommendations.”
His theory is that these recommendations and bundles play a vital role for the
“We did our own research and analysis of Amazon,” he continues. “We also had data from people who study them, and we found that the company drives 6-8% incremental revenue from bundling, which equates to £14bn per year. It’s quite a tidy sum.”
But we’ve skipped over the foundations of this start-up success story. So, let’s start at the beginning.
Finding cross-sales solutions
An entrepreneur with a background in chemical and computer engineering, Sri worked for Accenture in Internet and Strategy, before building and selling two digital marketing technology businesses. He then took up a scholarship to study Futuristic Tech and AI at the Singularity University, which is run by NASA. At some
“I kept thinking about how the on-site experience is inferior [to real life]. And that kind of led on to the cross-selling issue,” Sri explains.
“If you think about it, in the offline world, retailers do a far better job of cross-selling than those in the online world for a couple of reasons.
“In the physical world, retailers use principles of product relation and product promotion and the ability for a sales director to talk to customers. That enlarges the volume of cross-sales opportunities. Online, those are missed. When I reflected on it, I observed that there is no real cross-selling on websites.”
He explains that while there are recommendations on product pages, these take users to another page, where retailers may lose them.
“Now, most people don’t know this, but product recommendation was devised by Amazon to avoid site abandonment. However, it inadvertently drives cross-sales. We reflected on that and thought: ‘What can be done to improve a retailer’s ability to drive cross-sales?’ That’s when we observed the Amazon platform was using bundling and using machine learning to do it.
Product recommendation was devised by Amazon to avoid site abandonment
“The hypothesis was that if we can enable bundling at scale, via a software solution, retailers can do a better job at cross-selling without any extra effort. They can increase their revenue by 5-15%. That’s what we’ve been trying to do.”
Read more: The future of fashion
And now for the science bit…
Increasingly uses a form of artificial intelligence (AI) known as neural networking to generate bundles, as follows.
“Neural networking is a field of AI. The way to think about it is that AI has multiple areas within it. One field of AI is machine learning. And machine learning has different fields within it. One is deep learning, one is neural networking and so on… there are lots of different fields.
“Neural networking is very good at coming up with hypotheses based on supervised learning, which means that it’s not left to train itself. Then it predicts what the outcomes could be, and it does it with statistical significance and we want the predictions with high statistical significance.”
To apply this to an
“Neural networking is very effective for what we’re doing,” Sri states. “That’s why we’ve been seeing high revenue uplifts.”
In fact, the Increasingly website claims that it sees a 12% incremental revenue uplift, although Sri explains that this is now out of date and the figure is now closer to 15%.
Making cross-sales scalable
Of course, if cross-selling is driving these sorts of results, it begs the question: Why aren’t retailers already doing it? Sri is ready with an explanation:
“98% of retailers aren’t doing this and there are two reasons. One is that they totally could [cross-sell] with the solutions that already exist in the marketplace. However, it is highly manual. They’d have to say: ‘That product and that product go together’ throughout their product inventory. But then they’ll have problems if a product goes out of stock, or they have new products, or they don’t go together anymore.
“It becomes a nightmare. Clients say they’ve tried it once but never again. It just doesn’t scale for a retailer to do it manually.
It just doesn’t scale for a retailer to do it manually
“That’s where Increasingly comes in, using an approach that relies on neural networking to drive precision in which items are bought together.”
So, what’s the second part?
“Retailers need a platform that enables them to directly influence cross-selling. What does that mean? It means the ability to take any customer trigger and serve intelligent messages that encourage people to buy more products together.
“A trigger might be adding a product to the cart. The customer can configure the products in a modal overlay and add them to their cart without going to another page. They don’t have to browse.
“Another trigger might be that a chosen product is not available. An [overlay] appears, saying the product is not in stock but suggests the best alternatives. The customer can configure [the item] and add it to the basket right there.”
With such wide-reaching functionality, you may expect the installation of the technology to be a headache. However, Sri explains that it’s a simple three-step process.
“The first step is to give us your product feed. So, retailers have probably got a Google Shopping feed and that gives us the information on product titles, the descriptions and inventory. Secondly, they put our tags on their site – and that’s just a Google-Tag-Manager-type thing. And lastly, they share a one-time report of their last 6-12 months of
“We use that data to train the models, which means […] we expect to see results in days. Some clients have seen 6% incremental revenue in 5 days.”
Sri cites the case of one of Increasingly’s first clients, Travis Perkins. The builders’ merchants were one of the start-up’s first clients and launched on a performance-related pricing model, but the
“We drove 14% incremental revenue growth for them,” he says. “We’re now considered one of their five strategic pillars, which is really amazing considering that we only went live with the platform 19 months ago. They only signed up 18 months ago.”
He mentions that the company is now in talks with other DIY retailers, who will join their rapidly growing client database, which also includes REISS clothing and Canon.
Read more: Ecommerce glossary
Beyond the website
“It can also drive marketing behaviour,” Sri explains. “For example, take the concept of bundling for Google Shopping. It gives the retailer two bites of the cherry, not just one. You equip retailers to sell bundles, which Google supports, but nobody does because it involves lots of work. We can do it via automation, without any extra work.
“For instance, a customer might search Google for
“Where we get to with all of this is not just the on-site product, but the solution in marketing. It might touch email, Google Shopping, Amazon listings, display advertising and lastly in store.”
That’s right. It also presents opportunities on the high street. Sri highlights the example of a customer going in store to pick up click-and-collect parcels. Increasingly could allow shop assistants to recommend additional products at the point of sale. There has even been an instance of a client’s call-centre staff making use of the technology.
“It now represents 10% of their revenues, which is huge. The call centre staff are going to the product page that the customer is interested in and looking at our bundles and saying: ‘How about buying this thing with it?’
It now represents 10% of their revenues, which is huge
“We’re going to provide a system that is more robust for them, focusing on how the call centre can do a better job. Again, personalisation can help with that. [They can hone] in on their customer’s buying habits to recommend the right add-ons and it’s just going to be so potent.”
Going beyond brands
“As a consumer, you go to a marketplace because of many things, including the convenience of finding many retailers and products in one place. And we know that these marketplaces work. Etsy, Farfetch… they work.
“But what about flipping it on its head? Making any retailer also a marketplace in their own niche, without doing any work, by enabling cross-retailer bundling on-site?
“So, for example, you go to Samsung and although Samsung sell their cases and their accessories, how about selling every single thing out there in the world that would be compatible with a Samsung device? Samsung
But Increasingly can work around this, potentially allowing brands to stock related items from any company.
“That remains in the customer’s mind in the future much more powerfully, which therefore supports repeat custom. That’s the long-term roadmap.”
With so much potential, it is little surprise that Increasingly won eCommerce Product of the Year last year, and was runner-up this year in the Best Retail Tech of the Year category at Retail Week Tech.
“Forbes saw us when we were exhibiting at Retail Week Tech. and they were really excited about what we were trying to do. They ran a feature on us. It was
“We’re global, not just the UK […] we have circa 30 clients now, which is pretty damn good for 19 months and we’re profitable already, which is amazing. We have the lion’s share of clients in the UK
“We’re focusing on growing the machine learning models that we use and making ourselves more scalable.”
The future of retail
When asked about the wider retail industry and his predictions for future, Sri has a clear vision.
“There is a software solution, for instance, focused on improving the click-through rates of email marketing, using natural language processing. There are companies that just do that. They are focusing on a really big problem and using futuristic technology to solve it.
“This is what we’re trying to do too. Increasingly is solving a real pain point: how do I get more cross sales? The future is solving existing problems with cutting-edge technology.”
Read more: Top 27 Shopify apps
Over to you
What do you think is the future of impact upon the retail and ecommerce industries? Tell us your thoughts on Facebook, Twitter and Linkedin.