It’s a theme we keep coming back to at PreSeed Now:

How can ads be better targeted in a world after third-party tracking cookies, and where data protection is more strictly policed and brand safety is increasingly important?

Today’s startup has an answer that can make sure their videos get to as many of the right people as possible, to help them maximise their ad revenue.

Today we meet Contxt Media.

But first:

• University of Glasgow spinout Clyde Hydrogen has announced it has raised a £1 million pre-seed round for its technology, which “safely delivers high pressure green hydrogen both at scale and at low cost.”

• Do you know a startup we should feature here on PreSeed Now? Ask them to ✨ email me: [email protected] ✨ and tell us about themselves. We’re busy looking for startups to feature in May and beyond.

– Martin

This issue of PreSeed Now is brought to you by EHE Ventures.

Known for backing high-growth tech, EHE just announced they're building a £15m AI tech fund, and have finalised their core Fund Advisory team. The fund itself seeks to support and accelerate the growth of AI technologies, empowering the brightest minds in tech at pre-seed, seed and Series A.

Learn more about their mission and the fund here

Contxt Media wants to give media companies smarter video ad targeting

Third-party tracking cookies have a bad reputation, but they sure have been great for the ad market.

And arguably, tracking aside, relevant ads are generally a better experience for consumers than seeing the same generic placements as everyone else.

But as these cookies start to disappear, the search for new ways of targeting is front of mind for the ad industry. At the same time, brand safety is an important concern. Many brands don’t want to be seen next to Elon Musk’s politically ‘edgy’ posts on X, for example. 

Meanwhile, publishers are equally concerned about making sure their content attracts as many of the right kinds of advertisers as possible so they can boost their revenue.

Publishers working with video could have a solution to this problem in Contxt Media.

The startup works on the ‘supply side’ of the ad market. That means the content advertisers place their ads against. So customers for the tech could be broadcasters, streaming platforms, or even big YouTube channels looking to maximise their opportunities for ad revenue.

Contxt Media uses natural language processing and computer vision to analyse publishers’ videos. This analysis then powers contextual targeting, sentiment analysis, brand safety and suitability, and the ability to make sure content is rated correctly as suitable for the right audiences.

“In order for a publisher to truly understand their content, there's a lot of manual work that goes on by humans in verifying and tagging the content. So the ability to automate this is a huge cost saving for the publisher and it gives advertisers trust in what they're advertising against,” says founder Ed French.

How it works

Contxt Media can integrate with a publisher’s content management system. The tech transcribes the audio from videos and analyses the visual content.

Publishers can then tag their content with exact entities, topics, and sentiment. Then this information gets passed on into the ad server, so each video is categorised correctly.

“This allows video publishers to set up campaigns for specific advertisers, targeting their content with a high level of granularity… Publishers can understand content better, curate better media packages, and they can charge more and provide a better media buy for the advertisers.” says French.

As an example, French says the startup is doing work with music videos at the moment. At present, advertisers might block whole genres such as hip-hop, simply to play it safe, even though there are plenty of ‘clean’ hip-hop videos their brand would be fine to advertise against.

“That's where we come in. We process all of those genres, and then surface exactly what’s suitable to advertise against and not.

“Music videos are quite difficult to categorise. How suggestive is the dancing? How revealing are the clothes? That's why computer vision is so important, to be able to tag videos as suitable for the right age ranges.”

The story so far

French has a long history in adtech, through which he saw the rise of programmatic advertising and has followed its evolution to where it is today.

Most recently, he was the CEO of a company that provides speech-to-text transcription of TV shows for brands and broadcasters.

He’s the sole founder of Contxt Media, which he started work on last summer. After completing an MVP of the product in January this year, he says the startup is now signing up its first customers.

One place French sees a big opportunity is on YouTube. 

“There are no really good brand safety suitability tools on YouTube… there’s very sparse contextual data. There's very limited brand safety.

“This is where we want to sit, and this is where we're targeting right now; going through all the big YouTube publishers’ libraries, processing content, and going into their Google Ad Manager and then then tagging that with all the rich contextual data. 

“They can then sell their programmatic guaranteed deals with absolute brand safety, open up all their inventory, and create a lot more targeted packages.

“That's currently where we're focusing and where we're getting the most traction. Everyone we’re speaking to has a problem here. They know that they want to monetise better on YouTube. They know they want to be able to do contextual better on YouTube. And they can't rely on Google to offer that.”

Subscribe to Premium Membership to read the rest.