In recent months, there has been a massive influx of AI tools – so much so that there is now an AI arms race between the giants Google and Microsoft. This heightened activity is forcing companies to update and expedite their AI strategies. In a number of fields — including Customer Service, Sales and Marketing, Software and Data Processing, and Research and Development — the predominant idea is that you must increase the pace of operations, or risk falling behind your competitors.
With so many marketers asking questions about the challenges and opportunities surrounding AI, Luxid hosted a client webinar looking at how businesses can unlock the power of data and AI. The webinar was presented by two of Luxid’s leading authorities in AI, Ville Murtojärvi and Robert Nygård. In this post, we take you through some of the essential marketing key use cases for AI and give our expert view on how to get the most from the latest tools.
Luxid is well placed to provide an expert view as we’ve been a pioneer in artificial intelligence for b2b marketing. Five years ago, we were using AI in predictive applications, such as customer churn and lead conversion prediction. We were using powerful machine learning models to process large data sets to generate predictive insights. And we are now working with generative AI tools, such as ChatGPT, to leverage their massive potential in digital work, such as creative and coding.
To stay ahead of the AI curve, we’ve set up an AI Excellence Program, with AI champions monitoring the continual influx of new tools. We’re producing AI guidebooks – from basic level through to advanced – we’re attending the latest AI courses, and we’ve created an internal forum on Teams to share all our knowledge.
And it’s this extensive knowledge and experience enables us to give our expert view of the important use cases for AI. First, we will look at the top of the marketing funnel – creating awareness and interest and then into lead generation.
Driving efficiencies with the creation of buying personas
Using the example of a C-level person working in logistics, we have already been using a set of tools that can give us initial research information on online footprint, relevant websites, social channels, influencers and so on. This helps us give better input for the generative AI, as the more precise input you can give, the better the results. We were able to pull off pretty accurate and usable generic psychographics, demographics and even pretty good drafts of a buyer journey with our tests. Everything had to be validated, but it was a big time-saver on buyer persona creation.
Driving the effective usability of competitor research
At Luxid we’ve been conducting competitor research to drive the performance of client marketing for some time now. WIth our existing research tools, we could already see things like share of marketing voice, competition by region, traffic channel performance, and competitor social channel analysis. AI really adds to what we can assess: branding positioning and messaging, services, reach, reputation, content market and usage of calls to action. AI enables quick comparisons of the competitive data.
AI supporting the creation of concept imagery
Luxid uses image generating program Midjourney. We believe it’s the best quality out there at the moment, and it has a clear license for commercial use. Again, the success of out comes down to the quality of the prompting, be it image prompting – using an existing image to guide the AI in the right direction – or just creating accurate master prompts. Midjourney is very good, but it’s not there yet with accurate product imagery.
Boost serialised asset production
In digital marketing, you need to have a lot of variants of the core ad creative to avoid view “fatigue.” You can use master prompts to capture the look and feel of a brand and then create variants by just changing a couple of words in the prompts. AI works best when you don’t have to worry about getting a product 100% correct or have an actual person you need to recognise.
It is possible to create beautifully visualised QR codes using AI
It is also possible to train the AI program yourself to get better results. Some things are easier, for example, you can produce people's lookalikes with just one good-quality photo. With product imagery it is a lot trickier.
Overall, AI can make the planning and production processes much quicker and significantly reduce the use of costly stock materials. This then leaves more time for creative thinking and experimenting. It also raises the creative bar as you must beat the new base level set by AI to stand out.
As well as planning and production, there are also several important use cases in data & development.
Luxid’s new ChatLXD custom AI tool
Showing just how easy AI can be to use – when you know what you’re doing – our marketing data & analytics team created ChatLXD in just one day. We now have a smart, private chatbot attuned to Luxid’s knowledge. It uses Open AI’s GPT 3.5 turbo model as a base, which we’ve taught with Luxid’s website data. It’s an AI chatbot for Luxid’s customer service needs, which will provide more accurate and timely information based on Luxid data.
Using AI to enhance reporting
AI also has important implications for reporting and insights. You can start by getting all your data in one place, building dashboards, and then implementing AI solutions in your marketing and sales processes.
Regarding what this means for AI development in dashboards and reporting, Power BI and other business intelligence solutions are bringing prompts to BI tools so that the business user (Marketing Manager) can simply write what solution they need, and the tool will provide the dashboard, report or an analysis description. This frees up time for data analysts and scientists to work on more demanding analytical applications, such as AI proof of concepts needed for predictive marketing.
Safeguarding data privacy
We can see a growing number of really important use cases for AI in marketing. However, applying all the necessary safeguards to ensure data privacy is vital. Without the right approach and protocols, you could fall victim to confidentiality breaches, data misuse, and even AI hallucinations. There are some types of data you should never give the model; this includes personally identifiable information – or PII – which includes anything that can be used to identify a person’s identity.
Remember, AI tools store the inputs and prompts, so treat them in the same way you would when giving your data to another person. And don’t trust the AI’s output blindly, since it can have factual inaccuracies, or be based on outdated or biased information.
But don’t worry. Luxid is here to help. It could start as simply as an audit. We can map out the opportunities in your organisation to see where you could benefit from using AI tools. We can organise a use case workshop, typically split into a discovery training session, then an online workshop identifying specific use cases for your company. Or we could create an ‘AI in practice’ workbook with a roadmap for moving forward.