Brand safety has to happen before the ad serves. Block content at the source.
Fiona Salmon Managing Director, Mantis
Interviewed by Justin Cooke
Published
Fiona Salmon is Managing Director of Mantis, the contextual targeting and brand-safety platform built by Reach (a major UK publishing group). Her career runs across publishing, media sales, audience data, and ad tech, including senior roles at EMAP (now Bauer), FHM, OnePlusX (later Captify), and Reach. She joined media sales after noticing her flatmates in the industry were happier, worked fewer hours, and earned more than she did in PR. In this conversation she sets out the print-to-digital to programmatic to transparency story of the industry; the first-party data plus contextual at scale model; the block unsafe content at source so the ad can't appear next to it discipline; the over-60% of news has positive sentiment finding that argues against over-blocking; the advertising industry is hidden but enormous in carbon footprint sustainability argument; the efficient ecosystem is the sustainable ecosystem commercial case; the content theft by OpenAI is going too far warning for publishers; the Mantis is built by a publisher and that changes the perspective differentiator; the block-list technology is overhyped, all news is misread by the industry rapid-fire calls; and the closing observation: the publishers and the brands are the only people who really need to be in the ecosystem.
Complexity does not equal performance
On the lesson from the era.
Advertisers, agencies, and publishers have learned that complexity does not equal better performance. The ecosystem is still very complex; it's just more hops and more takers. We are moving to a healthier model, with fallout: MFA content (made-for-advertising) is being identified and eked out. Not where brands want to be. Shouldn't have been allowed in the ecosystem in the first place.
On real-time intent.
The loss of cookies and the move to real-time intent has brought contextual to the fore. Current brand-safety tech doesn't do enough of a good job. Today's contextual technology can tell a brand not only where people are but where they came from and where they went to. The targeting can be broader.
The performance discipline is moving from CTR toward outcomes-based KPIs. For me, the holy grail.
Over 60% of news has positive sentiment, and the over-blocking problem
On the gap.
News is still getting over-blocked. Not right for every brand to be there, but a whole audience is being de-targeted. Stagwell's research proves this out, and more brands are listening.
Working with several newspapers, we found that on a normal news week (not while the Iranian war was on), over 60% of news stories had positive sentiment. Brands are missing out on being next to the attention-driven, highly engaged audience that news content delivers.
On the right brand at the right moment.
Positive sentiment drives better outcomes. Negative sentiment can also drive better outcomes for the right brand: a charity targeting aid relief in Iran or Gaza placed next to a loss-of-life article will speak to a person's pocket. We've seen those ads' performance skyrocket. It comes down to brand messaging.
Smarter brand safety: block unsafe content at source
On the principle.
Smarter brand safety means technology that assesses the risk of the content at source and stops it making it into the mainstream so the ad can't appear next to it. Lots of publishers do this through editorial codes of conduct. On platforms, it isn't happening.
Meta and Google can build their own brand safety; they have many more engineers than we do. If they want technology that works, they can give us a shout.
On AI scale.
AI's biggest difference is scale. Contextual precision across many articles and videos is impossible by human review. Without AI, it's me and you watching videos and picking which are suitable; we can't do that at scale, but AI can. AI can be transparent if designed well: explain how it's derived, show how it works. Deployed badly, it adds opacity.
Mantis is built by a publisher, and that changes the perspective
On the differentiator.
An agency recently said they love that Mantis is the best contextual performer on plan (out of 11 vendors), and that by working with Mantis they're supporting the supply side and publishing. The agency had been told to spend less with ad-tech vendors. I would never have heard that before working in ad tech.
Mantis is not an additional tag in between the buyer and the supply. It's coming from the supply side, scanning the article or video content, deciding whether it's suitable for advertising or not. A video product is launching that does the same at scale using audio-visual signals (not metadata, not keyword data).
Efficiency is the commercial case for sustainability
On the fiduciary argument.
When I speak to a board or C-level, I talk about efficiency, not sustainability. The more efficient we are, the less data we transfer, the less AI we use (or use sparingly), the less carbon. Fewer hops between buy and sell, fewer tags. Underneath: a commercial differentiator. We're saving the planet and saving money. That's the business case.
On the gap in the ecosystem.
There's a gaping hole between demand and supply. Many of my roles have supported the supply side. What's needed: technology that helps both sides equally, rather than focusing on one. That's what Mantis is trying to do. Beneficial for both rather than concentrated on one.
The question for the board
If 60% of news carries positive sentiment, what share of brand-safety spend blocks at source versus over-blocks the high-attention journalism that follows?
