Nozzle – building an AI-powered business on the Amazon platform

 Our interviews with leaders of some of the most innovative, data-driven companies in the UK continues with this conversation with Rael Cline, Co-Founder and CEO of Nozzle.

For obvious reasons, ecommerce has taken a huge leap forward over the last 9 months. Amazon has been a major beneficiary of this with the company’s global headcount growing by 400,000 (28% growth) between January and October 2020 and revenue up 37% ($96.1 billion) for the third quarter of the year. Amazon, like Google’s search and advertising business, has seen an ecosystem of software developers and agencies emerge to support the merchants selling on its platform. Nozzle is using AI to help merchants and agencies optimise their sales on Amazon. The company recently raised $2.6 million in funding to expand their activities and Rael takes us through the opportunities and challenges of building a business around the Amazon platform.

 

Martin: Can you give a brief overview of what Nozzle does and why you and your co-founder set the company up?

Rael: Nozzle.ai is an eCommerce and advertising optimization platform for Amazon, headquartered in London. We’re a spin out of University College London’s (UCL’s) Computer Science Department that specialises in ecommerce and online advertising research. We combine top machine learning experts with proven experienced Amazon specialists to help brands and agencies drive the business metrics to succeed on Amazon, from sales or new to brand customers to profitability growth. We have a customer analytics product that gives insights into your customers on Amazon which we then tie into our Amazon advertising product that leverages those insights to acquire the right customers for your product.

Whilst Amazon has really granular raw datasets, they’re very siloed and need a lot of work to turn them into actionable data – a problem tailor made for AI to solve. 

 

Martin: Your core customer groups are marketing agencies and Amazon merchants. How do your offerings vary for these 2 types of customer?

Rael: We try to align these two customer types as much as possible. For instance, it’s not commonplace for large brands to grant agencies access to the retail (i.e. order) datasets. However this is essential to succeed on Amazon. We also try get alignment in terms of KPIs. For instance measuring performance purely on return on ad spend in isolation will ignore the ad’s contribution to organic sales (another reason to join the ad and retail datasets together!).

Both sets of customers typically want integrations with other software (dashboard software like Tableau or Microsoft’s Power BI for instance).

 

Martin: How does Nozzle use AI to deliver its services?

Rael: AI can be used in many areas. On the ad side it’s used to submit bids in the auction to show a search ad. There’s a lot that goes into the model as it’s a combination of the ad campaign goals (maximising for new-to-brand customers or a specific return on ad spend for example), whilst taking into account the budget, competition and other market-related trends. It’s very dynamic.

On the retail side, sales forecasting (and, therefore, inventory forecasting) is a huge pain point. If you run out of inventory on Amazon, you slip down the rankings which results in lost sales and if you have too much inventory that doesn’t sell, you will incur extra fees from Amazon if you’re using their warehouse to store your inventory.

 

Martin: How does placing targeted ads on the Amazon platform differ from using Google Adwords?

Rael: Conceptually they’re pretty similar; I search for something and there’s a real time auction to show me an ad in the search results[1]. When the page loads, some results will have a “sponsored” tag to differentiate from the organic ranking. Like Google, the details of the algorithm are not known exactly but we know what some of the most important variables are other than the actual bid price such as sales velocity, ratings, reviews, whether you have inventory available, etc. However, unlike Google, Amazon gives you more options to target a specific product instead of just a keyword. So I can target a competitor’s product for instance. Also, although not scientifically confirmed, successful Amazon search ad campaigns do seem to help the organic rankings too, unlike Google.

 

Martin: How does Nozzle use data to improve its products/services?

Rael: We use market data, usually broken down by category and country (i.e. electronics in the UK) to help guide our some of the algorithms outlined above.

 

Martin: What are some of the key challenges in providing complementary services to a massively powerful platform like Amazon?

Rael: There’s always the platform risk, i.e. Amazon can change the rules of the game unilaterally. However we have a really good relationship with various Amazon teams. Amazon has been and will continue to invest in their APIs which enables third parties like us to provide services on top. Ultimately, Amazon can’t be all things to all people. They’ve correctly focused on consumers so it’s really easy for us to search and buy something in 10 seconds on Amazon but for the entity on the other side of that transaction (the merchant), there’s a lot more friction. So enabling third parties to reduce friction and grow the merchant base feeds their flywheel. We don’t see that changing for the foreseeable future.

Of course, the regulatory climate is changing but we have always assumed the strictest interpretation of data privacy rules and regulations when building our product.

 

Martin: Do you have plans to extend your products to other platforms?

Rael: We are very focused on Amazon for the moment – there’s so much to be done! However we are starting to look at where other marketplaces such as Walmart and eBay fit into our roadmap. One cannot ignore Shopify for instance so this is a conversation we will have with our customers in Q1 2021.

 

Martin: What is Nozzle’s pricing/revenue model?

Rael: For the ad bid management software, it’s priced as a percentage of the ad budget. For our customer analytics tool, it’s more of a software-as-a-service (SaaS) model where there’s an all-in monthly fee according to the number of monthly Amazon orders.

 

Martin: Nozzle is based in London. How do you think the environment for start-ups in London will change once COVID-19 has been dealt with?

Rael: That’s an interesting question. I know a lot has been said about how COVID-19 has sped up trends that were already happening, working from home being one of them (especially for technology companies). The truth is more complicated. I think building a culture, which is especially important in the early days, can be harder to do if people are not all in the same physical space at the same time. It’s even harder if some people are and some are not.

Technology clusters happen for a reason. Whilst Zoom and Teams can be helpful to accomplish a lot of tasks and collaboration remotely, there are some things it cannot do and for that reason, physical offices will remain relevant. They main change may be to suit the type of work you’re doing. For instance a separate quiet area vs a dedicated collaborative space instead of a big open plan office. I think companies won’t need to assume everyone coming in on the same day so they may get away with only 65% of the office space too.

 

Martin: You recently raised $2.6m in funding. How does Nozzle plan to use that investment?

Rael: The funding will be used to speed up our product development on the engineering and data science teams as well as investing in business development.

[1] Amazon also has a display advertising product which are the banner ads that you can see whilst not actually on Amazon.com).