The Growing-Up Files, Issue 4
Four versions of you in one week
On Monday, someone on your team posts about a trend in your sector, and it does well, better than anything you've put out in a month.
On Wednesday, a sales deck goes out with a value proposition somebody wrote at speed the night before the meeting.
On Thursday, the welcome email that's been running since 2023 lands in eleven new inboxes, describing a business with one product and two people in it.
And on Friday you sign off the graphic for a trade stand, where the line under your logo is the one your designer liked best.
The result is 4 different images of the same company put out into the world, with no consistency in sight.
And none of the search engines or LLMs answering people’s questions know where to put you.
Nobody decides to drift
When there's no agreed answer to who you're for and what you're for, every decision defaults to the fastest available signal.
The algorithm rewards a format, so you make more of that format
A trend rewards a tone, so you borrow the tone
A competitor's post looks like it's working, so you try your own version of it
Every one of those choices is small, cheap, and easy to defend on the day you make it. None of them is a mistake in isolation.
Then a year of them stacks up, and the business is being described in a way that's 89 miles from where it started, and from why you started it in the first place. Nobody notices, because nobody ever holds this month's output up next to the reason the company exists.
That's the adolescent brand doing its thing in slow motion, except this time the message hasn't just been left behind by the business; it's wandered off on its own.
None of this is an argument against trends. A trend is a perfectly good input; it just needs checking against the strategy. Where there's no strategy to check it against (or the strategy was last mulled over 3 years ago and now it’s sitting in a downloads folder), the trend wins by default, every single time.
And now something else is reading all of it
While that was happening, the way your buyers find suppliers changed. Now a whole bunch of them are asking their robot LLM of choice.
Fine, you think - we’ve got great SEO!
But, brace yourself, the robot isn’t focusing on your website.
Meltwater analysed 9.5 million AI citations across six engines and sixteen B2B categories. User-generated platforms accounted for 47.5% of the citations. Peer review sites like G2 and Capterra took another 15%. Company websites managed 18.7%.
So roughly four in five citations came from a source you don't own and can't edit. You can't go and fix the Reddit thread. You can't rewrite the review, or the comparison a practitioner posted, or the way a customer described you to a colleague in a LinkedIn comment.
What these systems appear to be doing is looking for corroboration beyond a brand's own claims. They surface most reliably on decision questions, the how do I choose, what should I compare, how do I avoid getting this wrong, who should I be talking to sort of questions, rather than on anything that reads as brand-led promotion.
(LinkedIn features heavily in this, by the way. It was top five for citations in fourteen of those sixteen B2B categories, and first in Marketing & Advertising. Three quarters of the LinkedIn citations came from individual people rather than company pages, and over half from accounts with fewer than ten thousand followers. Expertise beating reach is the most cheering thing I've read all year)
Now put the four assets from earlier back into that picture. When the sources don't agree with each other, there's nothing stable for LLMs to latch onto, and you become harder to describe than the company next door. So the company next door gets described instead.
A client of mine said this to me last year.
"We are not making it into their consideration set."
And now the consideration set has a new gatekeeper 🤖.
This is why I start with your customers
People assume the qualitative stuff, the customer interviews, is the soft part of a research process. The warm bit, where everyone sits around talking about feelings.
In reality, they're the most insightful and valuable thing I do. The reason we go to customers first isn't sentiment; it's that the words the independent sources will use are already in their mouths.
Every review, every comparison, every reply in a thread gets written in the language your customers already have for you. If you don't know what that language is, you're guessing, and your own copy ends up in a different dialect from the rest of the internet.
So we go and find it. And then the real job starts, which is making sure that everything you do control says the same thing, in the same words, so that all of it corroborates rather than contradicts.
What you end up with is one clear answer to who you're for and what you're for, in language your customers recognise, written down somewhere the whole team can use it. After that, the Monday post and the sales deck and the welcome email and the trade stand all say the same thing, because everyone's reaching for the same document instead of their own best guess.
And what they’re putting out into the world matches what your customers are putting out into the world…
And the robots know who you are, and when to shout about you.
So what to do now?
Here are two things you can do this week.
First, pull four things that went out this month from four different people or channels. Read only the opening two lines of each. Would a stranger say those were the same business?
Second, open ChatGPT, or Claude, or whatever your buyers are using, and ask it what your company does and who it's for. You'll find out two things at once: which version of you is out there, and where it got it from.
If the answers make you wince, that's useful information rather than a disaster. Book a call with me here, and we'll work out how to fix it.
The Growing-Up Files goes out once a month, on how a brand's messaging has to grow up as the business behind it does. Next month: your team isn't the problem, but you may have accidentally made them dependent on you.
