How to Write When a Machine is Listening: 4 Tips for Earnings Scripts
We all know that the age of AI is here. AI is making trades while listening to earnings calls, scouring regulatory filings and occasionally - sending short-lived volatility into stocks.
As you prepare for earnings – see our 4 tips for speaking the language of the AI platforms who are already listening in.
1. Avoid mixing good and bad news in one sentence. Keep them far apart.
Don’t do what Elon did on Tesla’s Q2 2022 earnings call:
"The past two years have been quite a few force majeures, and it’s been kind of supply chain hell for several years. Credit to our awesome Tesla supply chain team for overcoming insanely difficult challenges."
To the human ear, it’s a compelling, even triumphant, narrative, with Musk acknowledging the headwinds but talking about the team’s success in spite of them.
But that’s not how a machine reads it. AI models read a litany of negatives, offset by one instance of good news.
Instead, practice disambiguation by not discussing challenges the business overcame in the same sentence as positive results. Cluster negative news together, if possible, in a discrete section or paragraph. Deliver bad news once, rather than repeating, and make messages from the CEO unambiguously positive if the CEO is discussing strong results or performance.
Here’s what the message could have looked like in a form AI is less likely to misinterpret:
CEO: "We have an extraordinarily talented supply chain team who have exceeded expectations in this environment and consistently delivered excellent results."
CFO: "The past two years have seen quite a few force majeures, and it’s been kind of supply chain hell for several years."
2. When discussing bad news, be more clinical and avoid conventional jargon – including overused euphemisms.
Take a look at this earnings call opening:
"Our third quarter results reflect a continuation of widespread pressures across our industry. Product affordability remained challenging due to macro-factors stemming from broad inflation, climbing interest rates, and continued low consumer confidence. In addition, persistent and steep depreciation in pricing impacted wholesale values throughout the quarter. Despite these challenges, we continued to gain share as we executed our strategic plan."
Oof. No matter how you cut it, it’s not great news, but with this language, AI is likely to see it as more negative than the numbers alone would say.
Instead, don’t give a litany of the macroeconomic troubles impacting a company – Investors are very aware of all of that.
Avoid excessive use of the word "challenge", which is generally viewed by AI as negative. Even when it’s a "challenge" the company overcame, using the term in close proximity to a topic like supply chain or operational costs will reliably trigger negative sentiment in some popular AI models. Ditto, constraint, inflation, disruption, and uncertainty.
Here’s an alternative that makes the same points in a way AI won’t interpret as negatively as the first version:
"Our third quarter results reflect the continuation of widespread pressures across our industry. Customer demand was impacted by product affordability as consumers navigate the current economic climate, which also resulted in persistent, steep depreciation in our wholesale business throughout the quarter. Despite these challenges, we continued to gain share as we executed our strategic plan."
3. Repeat the good news and boast (a little).
AI responds so well to the repetition of positive news that it’s worth writing in a way that might sound boastful to a human audience (but don’t overdo it). The machines are triggered by certain topics, like EBITDA and Share Buybacks. They "like" certain words and phrases and score them as positive:
New (paired with revenue, customers, etc.)
Similarly, there are words they don’t "like", and they will ding your script for them:
Challenge (in all forms)
Increase (Paired with "Costs")
4. Review available sentiment scores of past earnings transcripts on platforms like FactSet and learn more about how to adapt earnings communications to AI driven tools.
FGS Global has tracked the rise of sophisticated AI models that are being used by a wide range of investment firms, from quant funds and high-volume hedge funds to institutional funds like Blackrock, Nomura Securities, and a host of others. These firms use AI to model ESG, aggregate analyst expectations of inflation and are central to business intelligence platforms like FactSet and Bloomberg.
The right AI driven earnings communications strategy will:
Improve FactSet Sentiment Scores
Incorporate best practices for AI audiences
Have custom style guides that help minimize unnecessary volatility from AI algorithms around earnings.
For more information, contact email@example.com.