Many of us are unsure as to which uses of artificial intelligence are actually useful, but often we are only looking at content generation and not beyond to the many genuinely useful tasks in marketing strategy AI can aid with. In this blog, we have put together a list of some uses outside simple content generation that are great for agencies and in-house teams pressed for resources and time.
Contents
Tone of Voice
Analysing your tone of voice on social media or other platforms is a valuable analysis. It allows you to judge your results against your brand values and professed aims, looking for areas of adjustment that will ensure your personality comes across as intended to audiences.
The difficulty for many comes from analysing tone of voice impartially, and not allowing preexisting notions of your brand to influence your results. AI can be used here to provide a preliminary and impartial overview – here’s how you’d do it:
- If you were analysing posts on X, for example, copying the raw text of all posts from a certain date range into a language model such as ChatGPT will provide the context for the AI to base its analysis on.
- Provide an instructive prompt – don’t just ask it to “analyse tone of voice”, try inputting your brand values and asking it to judge it against these, or input competitors’ posts in the same answer and ask it to compare the presence of specific brand values against them.
- Ask for specific examples from the content – providing this layer means the output is relying less on the AI to generate its response from thin air, and really uses it more to speed up the reading process. The judgements made here are fairly objective, and this analysis can be trusted.
Sentiment Analysis
Sentiment refers to an audience’s attitude towards a certain topic (divided into positive, neutral and negative), and is an important assessment for customer service teams, emerging brands and established products.
Saving your budget on sentiment analysis tools that aren’t always reliable doesn’t have to mean trawling posts manually, assessing each poster’s sentiment individually. Pasting posts directed at a certain account, using a product keyword or a hashtag and directing AI to assess user sentiment is a great way to gauge how your audience feels.
There may be some limitations on the number of characters a model will allow per prompt, but a large number of posts are still able to be inputted and this analysis is valuable. Some tips are:
- Ask the AI to separate results into positive, negative and neutral, but add that you would like a breakdown of each by amount/percentage.
- Be careful that if you receive a large amount of negative feedback, traditional sentiment analysis tools have trouble detecting sarcasm and therefore mark such posts typically as positive/neutral. From our experience, language models are far better at detecting this – include a warning to watch for it in your prompt, and be wary that this slight discrepancy could be present in your results.
- As always, ask for examples, but you can also ask for a list of consumer’s most common concerns, issues, and praises – use AI to race through these reading analyses that aren’t difficult but would take a human a long time to complete.
Competitor Analysis
A key point with these uses for AI is that anything you can do for yourself, you can also do for your competitors. Furthermore, you can ask it to complete these assessments in conjunction with your competitor’s content, comparing and contrasting your strategies from an impartial perspective.
It’s not just posts – you can paste entire websites in text into prompts to AI and ask it to analyse any aspect of the content. Want to improve your homepage? Why not input it along with those of your top competitors, ask it to look for gaps and possible missing keywords, and then conduct further research to back it up?
Here is a simple analysis of our homepage, under the prompt “Analyse SocialB’s homepage – what is our tone of voice? If you were a visitor, would you feel you have a clear picture? Here is the content”:
ChatGPT specifically can be quite charitable on a first response – a follow-up to this could be asking for a list of criticisms/improvements.
Consumer Personas
You may have already built customer personas for your business, or have a rough idea of what they would be. As in many cases, AI can provide an alternate and impartial perspective that may bring up some points or niches you missed.
Add services, website content, social posts, demographic data from analytics or socials and more and ask for some different audience personas for your services. Remember to prompt the AI for justifications, and perhaps some additions like which messaging can apply best to certain people.
Limitations
The rise of artificially generated content has muddied the value of AI models, creating an over-reliance on a tool that most would agree is not fully there yet. The uses above are far more reliable, using AI instead as a complementary tool that can draw from existing information and instantly create reliable analyses.
It is essential to take any AI analysis as a part of the puzzle: the specificity of prompts and providing information can do a lot to offset flaws, but do not rely on it solely for any decision-making or work presented to clients. Think of it as an outside advisor, instantly able to parse information and provide instant references and suggestions.
