Content discoverability

Q: What have you learnt about using AI for content discoverability?

A: There are many opportunities for using AI for content discoverability. Some key areas of focus have been personalised recommendations, AI optimisation for SEO, using keywords and URL structures to improve Google traffic and enriching content by interpreting analytics and data into visual formats.

Over the last six months, we have interviewed publishers around the world, from the BBC to the Hindustan Times and there are some clear commonalities between their approaches to AI for content discoverability. Generally, they have prioritised admin tasks like tagging and internal linking in order to tell search engines what stories are about and to encourage readers to consume more stories. AI has proven extremely effective in providing personalisations of what to read next, for example.

When it comes to generative AI, publishers may use this to optimise headlines or generate keywords for SEO optimisation but every publisher we spoke to still had a ‘human in the loop’.

One key learning is knowing that hallucinations can exist where AI is used so it’s working with that knowledge and building in guardrails and processes around those possibilities rather than ignoring or hoping there won’t be any!

Other key learnings are around tracking changes and outputs when using AI for content discoverability tasks and linking the data to analytics where possible. This allows you to see what’s working and then adapt the AI workflows to improve the quality.

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

A: AI has proven particularly effective in helping publishers optimise audience reach and engagement through data-driven content recommendations. Publishers increasingly aim to streamline workflows and improve content relevance by leveraging AI to analyse vast data sets instantly. This allows AI to provide targeted suggestions for SEO-optimised headlines, trending topics, related article links, and ideal publishing times — all customised to the publisher’s own data and audience patterns.

Generally speaking, journalists can end up spending considerable time manually tracking data trends, whether that’s through Google trends or looking for patterns in their own data to gauge audience interest.

One example might be in planning an event that happens every year - like Eurovision or Glastonbury - so the same analysis is carried out over and over again to answer questions like, ‘What were our top stories last year that are definitely worth doing again’ or, ‘What didn’t work and where can we better spend our time this year?’

AI has alleviated this burden by automating the process, allowing editorial teams to focus on in-depth content creation rather than constantly monitoring trends.

With AI, instead of spending days on pulling historic data, previous trends, competitor data and so on, this work can be done in a matter of minutes so that journalists are just left with a clear plan of what to publish and when to publish it in order to achieve the best results.

This AI-driven approach not only increases efficiency but has shown measurable improvements in audience growth and engagement, as content becomes more strategically aligned with audience interests.

This shift is also empowering journalists, giving them the time and tools to produce richer, more impactful journalism.

The learnings for publishers have highlighted the importance of adapting AI outputs to align with unique audience behaviour, as well as fine-tuning algorithms to ensure recommendations remain relevant amidst evolving readership trends. Continuous training of the algorithms is key as audience distribution changes so quickly. Identifying where potential growth can come from and making tweaks to optimise for these opportunities is key to staying ahead of the competition.

Three best practice top tips

1. Use AI to make life easier for your journalists. In my experience, there are very few people who became journalists so that they could spend their days adding topics and links to stories. There are so many different ways to integrate AI into a newsroom but by far the most effective is to solve real problems and allow journalists and editors to focus on what they do best. Speak to your staff and find out where the real pinch points are that could be automated. Solving real problems and creating efficiencies will make adoption much easier.
2. Focus on data quality and relevance. Garbage in, garbage out as the saying goes. If you build a headline optimisation tool based on generic data, you will end up with generic headlines. Maintain high standards for data used in AI training to ensure reliable, up-to-date insights that genuinely reflect your audience’s interests. It’s important that you use the most recently available data possible and that it’s continuously updated. You don’t want to find yourself optimising for an old Google algorithm or audience trend.
3. Use a combination of in-house technology and partnerships. Wherever you are on your AI journey, using a combination of in-house technology and partnerships with experienced companies can be quicker and easier, helping you to avoid common pitfalls and save money on big builds. Once you know what you want to achieve, work out whether it’s better to build or buy.

Bright Sites is a leading tech company producing innovative content creation systems for publishers including our flagship product, Flow, an AI-enabled publishing platform. Flow is a versatile, high-performance CMS with liveblogging, ecommerce and AI and automation workflows enabling publishers to take advantage of the latest advances in AI.

Email: brian.alford@brightsites.co.uk

Tel: 07719 019091

Website: www.brightsites.co.uk