Q] Can you take us through Silverpush’s journey, since its inception? What was your first big milestone?
This journey has two parts. One is from 2012 to 2015, when we were building something very different and solving a very different problem. 2012-2013 is when smartphone penetration started in India, and brands began building mobile apps and spending on advertising on mobile phones. They realised that they could be targeting the same user on both desktop and on phone. Unfortunately they couldn’t be sure if if there was any duplicity taking place. So, the first problem that we were solving was cross device identification of users to understand the overlap between phone and desktop spends. To achieve that, we used data sets from phones and desktops, and basically matched users. That was version one of our product.
But then we realized early on that getting access to user data was becoming a little challenging. So, we took a call to pivot into something where we don’t need user data. At first we started looking at ways to target users without using their data. Then around 2015-2016, when video consumption started exploding, we realised that building contextual targeting for videos can solve the problem for video content. That is our biggest early milestone – a problem that we foresaw, and which became a reality within a few years.
Q] Help us understand some of the work that you do, and your products and offerings.
Contextual advertising would mean that we’ll need to understand what’s in the video or the context of the video. A video has different things going on in different frames, and so we built AI for understanding content within the video. Let’s say, someone is drinking coffee in the video. That’s a context that a brand like Starbucks would be interested in.
One of our current products is Mirrors, our hero product. It’s basically understanding the context within the videos using AI, and then using it to show relevant ads. The second one is Parallels, which is very similar to Mirrors, but a little different in the sense that we identify key moments from other sources. So, for example, during a cricket match, a player hits a six. This could be a key moment when we could show the right creative on Digital. It is about identifying the context from other sources to show relevant ads, and is built around cookieless advertising. That is how we would identify the theme of our products.
Q] What category of brands use this kind of tech, and in which geographies?
All the top brands and agencies work with us in the countries we’re present in. We started with India, and thereafter tested the product in Indonesia, and it worked well. Eventually we expanded to Philippines, Thailand, Malaysia, Vietnam, and Singapore in Southeast Asia. Then came the Middle-East, and in Africa, Kenya and Nigeria were already our early markets. In the last three years we’ve expanded to the US and Canada, which have become big contributors to our revenue now. And this year, we’ve started operating in the UK and France.
Our biggest clients are some of the biggest brands – FMCG companies like Unilever, P&G, Nestle; QSR brands like Coke, Pepsi, McDonald’s; automobile brands like, Hyundai, Ford, Nissan, and luxury automobile brands like BMW; and finally some high fashion luxury brands like Louis Vuitton.
In India we have worked with LG, Samsung, Havell’s, Dyson, HUL, Perfetti, Coke, Pepsi, and automobiles like, Hyundai, BMW, and Nissan, to name a few.
Q] How do you measure the success of a particular campaign? Please share your biggest case study or your most successful brand story so far.
One interesting brand success story I can think of is the latest Spider-Man movie that came in. The brand partner Oreo wanted to show their ad whenever Spider-Man appeared on-screen across YouTube, Facebook, or other platforms, including user-generated content, and even on Facebook reels. The idea was to detect every Spider-Man appearance in the content, which included kids wearing Spider-Man costumes, and show an Oreo ad. This was an interesting case study that was done across multiple geographies.
If we talk about measuring the success of a particular campaign, one is the KPIs as predefined by the brand. Let’s take the same example – because it’s a CPG brand, a key metric for them was ‘view-through rate’ of their videos. This means – whether their videos are being watched enough when they are placing the ads. We benchmark this against their view-through rate for the rest of the campaign, ones that are not working through our platform. We checked whether we are 25, 30 or 50% better than the rest of the campaign or not. That would be the metric we would want to work on. At the end of the campaign, you always get that information from the brand, and with our technology, it is usually always 25 to 50% better.
Q] How has your company grown in the last few years?
From 2015 to now, we’ve been growing at a CAGR of about 70% Y-o-Y. Even this year we will be doing the same. In 2015, we had 7-8 cr in revenue when we started selling this product, and this year, we’ll be closing at about 200 cr. Next year we are looking to cross approximately 350 cr. This is across markets, and the India-market would be about 10% of it.
Q] Going ahead, how do you plan to continue harnessing the power of AI for your new and emerging customer base, and improve on brand growth and customer engagement strategies?
AI has been at the core of our products. When we first started doing video AI, we did not even have the infrastructure. So, we built our own hardware for processing video content via AI, and today AI is evolving. So, we are using it in multiple ways now. We are looking to incorporate generative AI and Chat GPT into our technology and our products, and offer additional value to customers.
Second, we define ourselves by how much more value we can offer in the cookieless space. If I look at the advertising industry as a whole, cookieless advertising is one of the most important things that’s happening. The other is CTV growth, which has already happened in the US, and is yet to happen in this part of the world; that’s another interesting space.
