Advertising

Q: What have you learnt about using AI in advertising?

A: Over the past decade in AI and advertising tech, we’ve learned that AI excels best in areas like ad serving, proof-reading, slogan and button suggestions on banners, and lead discovery.

However, AI should always act as a co-pilot, not work entirely on its own. Despite good prompts, AI can still hallucinate or make mistakes in around 10% of cases, requiring human oversight.

For ad creation, while AI can generate 100 banner variations, the quality of 90% remains subpar, especially with AI-generated images. It’s often more efficient to start with galleries of ads that have worked well in the past and use these best-practice ads as a template, using AI for minor adjustments or slight variations like different backgrounds, products, button texts, etc.

AI excels at image enhancements, such as extending photos, applying filters, and cropping elements, but it still falls short when it comes to generating images from scratch. The expectation that advertisers would never need to provide their own images hasn’t been met so far. Instead, the starting point must always be the advertiser’s photos, which AI can then enhance or multiply.

Sales automation, particularly for smaller advertisers, shows promise but remains underdeveloped; automated sequences often come across as too generic (“... love your company...”) or inaccurate, which can damage advertiser relationships. Fixing these issues can be time-consuming and negate any time saved. However, AI can be highly effective in identifying triggers for human outreach, such as when an advertiser shares a social media post that could be converted into an ad with the publisher. These triggers can also involve aligning the publisher’s topics and products with the content on the advertiser’s website and social media profiles. The focus should be less on full automation and more on optimising lead lists and prioritisation.

Additionally, we’re noticing prompts becoming more sophisticated — structured in sequences rather than simply longer, utilising different models at various stages of the process. To handle these advanced AI cascades effectively, publishers require specialised partners, as managing and optimising this infrastructure is becoming more complex and costly. This is especially true given the 10x or 20x cost differences between certain models, where not every prompt requires the most expensive option, while others do. Navigating this intricate landscape demands time, tools, and focus.

Publishers should avoid relying on a single AI model to determine if a business case can be automated or optimised; instead, experimenting with multiple models is essential before drawing conclusions.

Q: In which use-case has AI proved most effective?

A: We observed a successful case where a publisher created a comprehensive company directory of all local businesses in their region by intelligently aggregating / crawling data from various sources to deliver a top-quality product for readers and Google by letting AI write great company profiles based on the data input.

Initially, they offered directory listings as a complimentary feature for a limited period, using the data to generate strong conversation starters with local businesses. This information was then also utilised to build more accurate and up-to-date lead lists for sales teams.

For digital sales teams, the new business directory provided an excellent opportunity to reconnect with the 80% of local advertisers who had stopped advertising, showcasing the publisher’s capabilities and reopening discussions. Print sales teams bundled the company listings into their ads and upgraded the perceived value of a ‘print ad’ massively.

Three best practice top tips

1. Partner with a specialist in ‘AI for advertising sales’ (not just general AI) who iterates quickly and is committed to results by offering a revenue share model rather than charging by tokens, ensuring shared success from new business generated.
2. Leverage AI to deliver an exceptional product for readers and surprise your advertisers with something innovative, free, or ‘out of the box’ — something made possible by AI, building on top your core strengths to re-engage advertisers to talk to you again — or more often.
3. AI can be a game changer in sales when you deliver services to the advertiser that almost cost you nothing (thanks to AI) but you deliver services and results before asking for the advertiser’s money. Reverse the usual order by offering potential clients free previews, spec ads, and more upfront, creating impactful surprises before closing a deal.

Smartico supports news publishers, boosting ad revenue through its AI-powered advertising suite, enabling the creation and management of effective local ads and campaigns at scale. With a success-based pricing model, Smartico operates over 500 advertising profit centres for media companies such as Axel Springer, Mediahuis, and The Irish Times.

Email: christian@smartico.one

Website: www.smartico.one