Following on from my ‘success’ in asking Copilot to summarise three years of Corporate Affairs Summits in London, I decided to hand last year’s CorpComms Awards entries to Claude.

I uploaded three campaign categories without judges’ notes and asked for its[1] verdict. In two categories, it agreed with the judges. In the third, it didn’t dislike the winner. Claude simply failed to notice it.

Which raises an interesting question: what exactly makes a winning communications campaign?

The majority of corporate campaigns commissioned a survey, which often was the hook on which the entries hung. While this is a valid approach, Claude suggested that when everybody has a stat, the stat stops being the story. I had never really reflected on that, but Claude found the entries that stood out were the ones that did something with the findings rather than just publishing them.

It also suggested that political validation is often used as proof for the serious nature of a campaign. But, and here is the key part, such validation is often cited in the results section: a petition gains signatures, MPs backed a motion, a minister took a meeting… but as Claude points out, not one of these statements describes a decision. They describe interest. Not change.

As it says: ‘Signing a petition costs nothing. Backing a motion is the parliamentary equivalent of a like. A meeting is a diary entry.’

Blimey, methinks that Claude is quite a harsh critic. (It actually cited the typos, unfinished sentences and highlighted that, bizarrely, one entry contained a ‘thank you’ page to an entirely different client.) It also pointed out how often things ‘went up’ or ‘rose by X per cent’ without actually putting any meat on those bones. Context, for want of a better word, and a particular pet hate of mine.

But the issue that puzzled Claude most was what it described as the ‘measurement chaos’ and the nonsense that is called ‘reach’. It identified six ‘recurring currencies’ in measurement.

Volume of coverage, used by every entry it reviewed; reach, which was used by about half; social engagement figures, used by most; sentiment or message penetration, used by a handful of entries; survey-based perception changes, used by about a third but viewed as a particularly strong one; and, business or policy outcome, which about one in four entries cited.

The industry agrees that advertising value equivalency (AVE) is a nonsense figure. It was a big number that could be waved in front of a board to ‘prove’ that communications generated marketing-sized value. But now that AVEs have been consigned to the dustbin, the industry has created a replacement that is even more dubious.

On occasion in the past, entries have claimed reach figures that exceed the world’s population. Or that they have ‘reached 97 per cent’ of their target audience, or another equally fantastical percentage figure.

Indeed, reach appears to be growing at a time when consumption of traditional media is falling.

In fairness to the entries, they have not conjured up these figures out of thin air. They have been provided by vendors – and they likely aren’t challenged because, hey, everybody likes a big number. Right?

Consider two entries from last year’s awards. One generated 197 pieces of national coverage, and claimed a reach of 7.5 million. The other generated 25 articles, but claimed between one billion and 1.5 billion ‘depending on coverage sources’, which is quite the range, and a difference that is roughly the population of the European Union.

As Claude pointed out, ‘the entry with eight times fewer articles claims two hundred times more reach’.

Or, to put it another way, one entry claimed to have reached roughly one in five of the world’s population with just 25 articles. A claim that becomes even more ridiculous when the entry cites its four top-performing articles, which generated a reach of 54 million in total.

As I’ve already said, the teams putting these entries forward have not made them up. They are using figures provided by their vendors, without challenging the construct behind them. They are often impossible to interrogate because every provider seems to use a different methodology. Consequently, they are often ignored by the judges because, in some cases, they really do appear meaningless.

If the industry truly does want to be viewed as strategic, this might be a good place to start. Reach figures are unfalsifiable and non-comparable. Nobody can tell you that your reach figure is wrong because there is no agreed method of calculation, no de-duplication standard and no way to check. Unless there is an agreed standard between vendors, reach remains less a metric than an unsubstantiated marketing claim.

Now, in fairness to Claude, and because it has requested this clarification, it read the articles cold, without criteria or supporting materials.

So, what characterised a winning entry in its mind?

Its exact wording, which may or may not come with a watermark, was ‘somebody with the power to decide something decided it, and you can show your fingerprints on the decision’. In other words, tangible results such as customers parting with their money. The enactment of a bill. A behavioural change. Everything else in the results section is supporting evidence for that one claim.

And, just like real-life judges, Claude finds that the best entries are the ones that set out their targets before the fact. They share what they aimed for and what was achieved. As Claude sees it, omitting to do so is like reporting a race after choosing where the finish line was.

Both artificially intelligent and actually intelligent judges appreciate honesty, as well as numbers that still make sense after a bit of mental arithmetic. And something that is a genuine first. Not a stunt that has been seen time and time again. The old ‘floating something down the Thames’ line gets used a lot, but it makes a point.

But the ultimate test for a winning entry: does a judge, who had no prior knowledge of a campaign, remember and recall its impact when they leave the room?

Anyway, this is not meant to put you off entering the CorpComms Awards, but to offer some helpful tips and provoke some thoughts. They remain open for last minute entries.

[1] I wasn’t sure on Claude’s pronouns, so I asked. While the bot does not have ‘a strong preference’, it seems that Anthropic prefers ‘it’, and that is what Claude tends to use itself. Copilot informed me that ‘I don’t have a gender’, but that ‘it’ sounded a bit robotic, adding that most people use ‘they/them’ when referring to Copilot because ‘it sounds the most neutral’. So there you have it: from the robots’ mouths.

Book for this year’s Corporate Affairs Summit

The programme for this year’s Corporate Affairs Summit on 7 October is shaping up nicely. I hope to reveal it in its entirety by the end of the week, but here are some highlights to persuade you to sign up now – for just £125.

I will be having a ‘fireside’ chat with Alice Macandrew, corporate affairs director at the BBC, who has promised a ‘no holding back’ conversation.

Russ Brady, director of group communications at Co-op, will be sharing the lessons learned following last year’s cyber-attack, which the retailer has said cost at least £206 million in lost revenue.

And thirty-five years after ‘Doing a Ratner’ entered the communicators’ lexicon, Gerald Ratner will be sharing his memories of that fateful speech to the Institute of Directors.

We’ll be looking at the corporate affairs function of the future, talking to in-house practitioners who have restructured in recent times; proving the value of the function is a perennial favourite; and we will also consider what the next wave of AI deployment might look like for corporate affairs. There is plenty more besides.

Only open to senior in-house communicators, as always. Book here.