How AI reveals which companies tell a clear story

Have you heard of the Thud?

According to Brett Miller, head of data, analytics and AI at Sodali & Co, the Thud is one of the simplest ways to determine whether a company has a clear story to tell.

Simply print out the transcript of an earnings call, pick it up and drop it on a desk. The louder the thud, the more you should bet against the company.

Miller, who previously worked at BlackRock, the world’s largest asset manager, describes the metric as ‘one of the best performing signals I ever created’.

It sounds frivolous, but the insight is serious. ‘The longer your story is, the more surface area you have, and the less clear and direct you are,’ he told delegates at this year’s Corporate Affairs Summit during a conversation with Sodali’s senior managing director Victoria Palmer-Moore.

As Miller put it, the numbers in an earnings announcement rarely surprise investors because ‘capital markets have become incredibly efficient at pricing in what a company is going to do on the top and bottom line’.

So, when a share price moves sharply after results, it’s often because investors have detected a different story altogether: a shift in strategy, a change in positioning or even a new leader.

And now investors have a new way of analysing that story. Using AI, they can interrogate a company’s narrative with the same rigour that they once reserved for its accounts.

So, what do they look for?

It’s bad news if you fancy yourself as something of a wordsmith. Investors count the number of adjectives and descriptive words you’re using and will compare that with the number of formal figures. ‘You should absolutely use adjectives. It gives colour and that’s valuable,’ said Miller.

‘But if you're seeing 11 adjectives for every one number being offered, it might be an indication for investors that they're overselling.’

Luckily, individual words matter less than you might think, because AI allows investors to capture the meaning behind them. ‘They’re looking for hedging; they’re looking for a lack of confidence,’ said Miller. ‘One of the best performing signals that you will ever find is just the quantified outlook ratio.’

In other words, when it comes to sentences about the future, how many contain a number? As he put it, ‘we expect tremendous growth’ or ‘we’re very optimistic about the future’ are not the same as ‘we expect 20 per cent growth’. Vague language, Miller opined, is what people use when they don’t know the answer, don’t like the answer or have simply been told not to answer.

Another indicator is what Miller calls the ‘resilience contrast ratio’, or, as it’s known more informally, ‘good house, bad neighbourhood’ language.

How many times has a CEO said: ‘We had a strong quarter despite a challenging environment’? Investors’ tools pick up on the sentiment, not the words, and the underlying sentiment is that times are hard. And that’s also what ends up in the analysts’ notes.

As Miller said, you don’t sell shares by talking about what a difficult time it has been. ‘Have any of you ever heard a realtor say This is a challenging neighbourhood, but the house is great?’ he asked. ‘If you beat, you beat!’

As for burying bad news… the more often a reader is sent away from a statement to Appendix A, say, and the further they have to go, perhaps Appendix Q, the more suspicious they become because, Miller argued, ‘there's a tendency for that to correlate very strongly with the area that the company wants to avoid the most’.

Indeed, the more a company makes an investor work hard to get the full story, the more they cut out the intermediate steps and go straight to where the horrors usually lie… the appendices. Miller likened it to a doctor pressing on a sore arm; they’ll keep pressing until they find out where it hurts. ‘The eye movements and the footnotes tend to be for companies where people know it hurts the most,’ he said.

But results are only part of the story. Investors now build a continuous timeline of everything a company says and look for any contradictions. Miller pointed to tech CEOs who declare engineering is dead, while roughly half their job adverts are for software engineers. ‘Engineering can’t be dead, and you’re also hiring for these roles,’ he argued.

And then there’s the Holy Grail of information for investors: a CEO opining on a podcast ‘because for whatever reason, they think the rules don’t apply’.

Miller's biggest point, though, was less about adjectives and word patterns but about the relationship between reputation and value.

Corporate affairs and investor relations are often treated as separate disciplines, with CEOs more deeply involved in discussions about valuation than reputation. Miller argued that distinction no longer makes sense because the two increasingly drive each other.

Investors today analyse far more than financial performance alone. In fact, they go searching for clues in places where companies might assume they would never look. They might investigate relationships with employees, regulators, NGOs, activists and other stakeholders because these could generate information about a company’s wellbeing.

It might be an employee with a public gripe about working conditions, a tricky relationship with a regulator or a reputational issue simmering away in the background… each can affect how a company is valued by both public and private markets.

Communications are now fundamentally data in a way that they have never been viewed before.

Or, as Miller stated: 'Your reputation is your valuation.'

Postscript

Having spent the best part of 45 minutes putting the fear of God into communicators, Miller ended on an optimistic note. A ‘what’s good for the goose is good for the gander’ piece of advice, as it were.

AI, he argued, cuts both ways.

‘You are being studied by a million different stakeholders, but AI gives you the capability to study those stakeholders just as closely so that you can find opportunities to really land a message when it is most impactful for the audiences that you care about.’

Rather than focusing on the threats, he said, communicators should go on the offensive.

Or, as Miller put it: ‘Use all of the data that exists in the world and identify the patterns that allow you to maximise your voice and your messaging.’