Chapter 13

Be known

You're never just known. You're known for something, at a moment.

My favourite from-zero story is my second agency, Cru Digital. It's probably the one people still remember.

It started with two of us: my close friend Damien Bredberg, a photographer, and me, a web designer. It was July 2004, and I was working full-time at Ballistix, a sales process agency. Great company; I was sad to leave. But a bunch of Damien's photography clients wanted websites, and I was itching to use everything I'd just learned about marketing and sales process. So off we went.

Damien's clients gave us a few months of runway at most. We had to bootstrap hard.

Ballistix had taught me the power of a good client newsletter, so that's where we started. We paid a freelance designer to create a set of postcards, and they really were cool. On the back was an offer to sign up for our newsletter. We dropped tens of thousands of them into local business PO boxes. It cost something ridiculous, maybe twenty cents a card, delivered.

About a thousand people signed up, nearly all of them business owners. Then I went to work. I wrote the most compelling, best-designed newsletters I could, packed with insights, marketing ideas, competitions, anything I thought an owner would actually want to read.

For the first few months the work only trickled in. It was barely enough to live on, even with no running costs at all. But I kept the newsletter going, every month, no exceptions.

About three months in, the trickle became a stream. We were busy. We hired two designers and a web designer to keep up.

Within a year we'd started on Google AdWords and moved out of the home office into a fairly big space in West End, Brisbane. We had about fifteen people by then, and twenty-five within eighteen months. We ran our own local events and turned up to community and industry ones. We ended up on just about every ad agency's supplier list for websites.

For a few years Cru was the go-to digital agency in Brisbane, known for the most creative, most immersive websites around. We won more than a hundred international awards over about five years, before we merged with Junior to form JuniorCru.

What it taught me is that you can manufacture awareness, if you're committed enough. And you can go from nothing to something substantial very quickly.

One caveat. I was in my early to mid twenties, and I routinely pulled all-nighters and worked eighty-hour weeks. Once I became a father, I went looking for a quieter agency life, and I've kept Evolut deliberately small to keep some balance. I haven't always managed it.

Known is the first gate. A buyer can't choose a brand they can't bring to mind, and neither can a machine.

Most brands measure it with one number: awareness. That number is useful, but it says nothing about which buying situations bring your name to mind, and which leave it out.

So this chapter asks a narrower question, twice, of the buyer and of the machine. When this buyer is in this moment, does your name come up?

Memory is filed under moments

Buyers don't search their memory by brand. They search it by situation.

Professor Jenni Romaniuk of the Ehrenberg-Bass Institute calls them category entry points. They're the cues buyers use to reach into memory when a need arises (Ehrenberg-Bass). This book calls each one a moment. Hunger at four in the afternoon is one. A letter from the tax office is another.

Moments matter after the sale too. Romaniuk surveyed US business-insurance decision-makers for LinkedIn's B2B Institute (Romaniuk). The more buying situations customers linked to their insurer, the less likely they said they were to leave, in 16 of 17 products.

The report's reading: "for each 2 additional CEP… probability of defection lowers by 5%". That's stated intention, not actual switching.

A single awareness number misses this. Ehrenberg-Bass researchers compared prompted awareness with the share of buyers who linked a brand to buying situations. They found "high stability and little congruence" between the two, over time and at brand level (Ludwichowska and colleagues). The awareness number isn't wrong. It hides where you're remembered, and where you aren't.

Fame breaks ties

Recognition still counts. In a classic experiment, "brand awareness was a dominant choice heuristic" (Hoyer and Brown). People picked the brand they knew, even when it was the lower-quality one.

Machines do the same. Xi Chu and Yupeng Hou had three AI models choose between one real skincare brand and nine invented ones, all with identical specifications. The models picked the real brand in all 670 valid trials (Chu and Hou).

Then a rival got a slightly better rating. A lead of 0.075 stars wiped out the famous brand's advantage. Once quality information was present, brand explained just 1.2% of how products were ranked.

So in this set-up, fame was a tie-breaker. It decided when nothing else did, and gave way to a small, real difference. The rivals were invented, though, and in the wild fame may well do more than break ties.

Either way, a tie-breaker needs a tie. To be in the tie at all, the machine first has to retrieve you for the situation the buyer described.

The machine knows you by what's written and searched about you

The best evidence on how machines retrieve brands comes from Edward Malthouse, a Northwestern marketing professor, and his colleagues (Malthouse and colleagues). They asked six AI models for 2,400 lists of recommendations in five categories. In the 1,200 lists from plain category prompts, many established brands never appeared.

L.L.Bean and REI were missing for hiking jackets, Craftsman and Black+Decker for drills, Braun and Philips for coffee makers.

Then the researchers described a need instead of naming a category. Craftsman rose from 0% to 35.4% of lists. L.L.Bean rose from 0% to 5.4%. With prompts built on each brand's own positioning, the figures reached 81.3% and 88.5%.

The machine knew these brands all along. It had filed them under particular needs, and the plain category question didn't open the file. The question, in the authors' words, is whether a brand's representation "is activated when consumers express needs the brand is intended to serve." The authors flag those probes as illustrative and few, but it's the closest evidence I've found that category entry points apply to machines too.

Across 82 brands, Google search interest was the strongest predictor of how prominently a brand was recommended. News mentions also counted. Advertising spend didn't. But fame as brand surveys measure it showed "little systematic relationship": Disney Cruise Line scored highly for salience in the BrandZ survey and was almost never recommended. The authors stress that this analysis is exploratory and doesn't show cause.

Vendor data points the same way. Ahrefs compared 75,000 brands against how often Google's AI Overviews showed them (Ahrefs). Mentions of the brand across the web correlated at 0.664. Branded search volume came in at 0.392, and ad traffic at 0.216. Ahrefs adds its own warning: correlation isn't causation.

So the machine knows you by what's written and searched about you, far more than by what you pay to say.

Two memories, one map

There are two memories to win, the buyer's and the machine's, and they often disagree. A firm with loyal local customers may live in buyers' heads and nowhere in what's written.

VOCAL shows it. Two former Weber Shandwick staff founded the Sydney PR agency in 2026. They asked ChatGPT and Gemini to recommend a firm in their city, and their agency was "absolutely nowhere to be seen" (B&T).

The founders were known in the trade, but the new firm had little written about it. Their plan is to build "earned authority" through other people's advocacy; there are no results yet.

Lay the two memories over your map of moments, and each moment falls into one of four states. The states are my own way of sorting the work.

An unwritten moment needs coverage, reviews and talk where the machine reads: PR, review and community work, with a named owner, in-house or agency. An unfelt moment needs reach and repetition among buyers.

A few moments, not one

The usual advice is to pick the one thing you want to be known for. Ehrenberg-Bass argues for breadth. In Romaniuk's words, "Brands with wider, fresher CEP networks have higher acquisition… also have higher retention" (Romaniuk). The insurance result points the same way.

So my working rule is a few chosen moments. Start with the one that matters most, win it in both memories, then widen.

One caution. Bigger brands get linked to more situations simply because more people buy them. The useful signal is whether you're linked more often than your size predicts. Romaniuk's work on Dirichlet norms sets out how to judge that (Romaniuk).

What ties a name to a moment

Snickers tied itself to a moment, hunger, with "You're not you when you're hungry". Its awards paper reports sales up 15.9% in a year, share growth in 56 of 58 markets and global sales up US$376.3 million (IPA).

The meerkat shows the other half: a distinctive way of being remembered. In 2009 a British comparison site, hard to tell from its rivals, launched a Russian meerkat called Aleksandr Orlov (AdNews). More than fifteen years on, the campaign still runs in Britain and Australia. It didn't describe the product better. It made the name hard to forget.

A brand needs both halves: a moment to be recalled in, and a cue that makes sure it's noticed. And once a brand is the name people ask for, it stops waiting to be recommended, as Booking's direct bookings show (Chapter 3).

How to tell it is failing

• Ask customers what was happening when they first thought of you, and the answers scatter. None match the moments you market to.
• AI assistants describe you accurately when named, but never name you when a buyer describes a need.
• Your marketing promotes a different moment every quarter.

The method: the moment memory map

The moment memory map checks each chosen moment in both memories: buyers' and the machine's.

1. Choose the moments. Take three to five from your Chapter 12 map, written in the buyer's own words. Romaniuk suggests prompting with the W's: why, when, where, while, with or for whom, with what, and how they feel.
2. Rank them by the 3C's. Romaniuk's report also offers three tests. Credibility: can you claim this moment believably? Competition: who holds it now? Commonness: how often does it happen?
3. Ask customers what was happening. Interview recent customers, and code the answers with the guide below. Count how many name each moment unprompted.
4. Read your branded search. Look for your name searched alongside a moment's words. This is my working rule for a cheap proxy, not a tested measure.
5. Ask the machine three ways. Malthouse and colleagues used three tiers of prompt: category only, needs-based, and positioning. The gap between them separates "the machine does not know us" from "it knows us but is not cued".
6. Sample properly. Fix the list of competitors before you start. Run each prompt many times and record the share of runs that name you. The researchers used 40 runs per model; an owner can start with five.
7. Ask the machine about you by name. "What is this brand known for? When would someone choose it?" Compare the moments it gives with the ones you chose.
8. Mark each moment's state, and assign the work. Held, unwritten, unfelt or open. Repeat it each quarter with the same prompts.

Sampling matters. A remodeler once showed Marcus Sheridan, the marketing author, a ChatGPT answer naming it "one of the best remodelers in our city". The same question on his account didn't (21 Hats). One answer is one draw.

Rate each memory on a three-point scale.

This scale is my own working rule, not tested against sales. Strong in both is held. Strong for buyers and weak or absent for the machine is unwritten, and the reverse is unfelt. Weak or absent in both is open.

Ask what was happening

The interview is the cheapest part of the map, and the part most teams skip. "Why did you choose us?" gets you a reason. "What was happening when you first thought of us?" gets you the moment. "Our accountant said the books were a mess" is a moment. "You seemed good" isn't, so ask what happened just before. Ten interviews is enough to start. Code each answer on five lines:

• Cue: what brought the need to mind, such as a letter, a date or a bad month.
• Trigger: what turned the need into action that week.
• Person: who was involved, or who they asked.
• Words: their phrase for the problem, written down exactly.
• Moment: cue, person and words combined into one line. Count matching lines across interviews.

The bookkeeper's moment memory map

The bookkeeper for cafés is an invented example: "Bookkeeping for cafés. Your wage cost against takings, every Monday by nine." She interviewed ten clients and ran needs-based prompts for four moments.

Her strongest moments live in people's heads, but almost none of it is written where a machine would read it. The machine knows her for a set-up guide she once posted, which isn't her chosen moment. Her work this quarter is to get others writing about wage cost, not to publish more of her own.

Fieldline's moment memory map

Fieldline is an invented software firm of 200 people. It sells scheduling software to field-service teams: plumbing, heating and cooling, and electrical contractors. A committee buys it: operations, finance and IT. Its signature claim is cutting technicians' travel time, based on a 2023 pilot.

From here to Chapter 19, one invented deal runs through every stage. The buyer is Northgate Heating & Cooling, a contractor with 80 technicians. Dana, head of operations, wants the change and champions it. Ravi leads finance. Mei is the IT manager. The owner, who is also chief executive, signs.

Dana remembers Fieldline from a trade talk on travel time. Ravi and Mei have never heard of it.

Fieldline is known for one moment, by one person. That gets it onto Dana's list, but each person in a committee arrives with a moment of their own. Its next quarter's work is a job-costing story, told by a customer's finance lead, where finance leads read.

By mode

What the memory is tied to changes with the mode. Habitual buying hangs on an everyday cue, so build broad reach repeating one cue. A quick choice hangs on the moment of need itself, so be present where the need strikes. Considered personal buying starts with life events; owner-led buying with the owner's worry about cost, risk or time. A committee brings a different moment for each person, so earn one moment per role, in the places each role reads.

How Known connects to the other stages

The moments come from your map, and distinctiveness, from your reason, gets you remembered. Known feeds Found: the machine can't find you for a moment it hasn't filed you under. And it shares its sources with Believed, because the coverage others write builds the machine's memory too.

What not to do

• Don't trust a single answer. Measure a share of many runs.
• Don't try to buy the machine's memory. In the best study so far, ad spend didn't predict AI prominence.
• Don't read a big count as a win. Big brands get linked to more moments because they're big.

How you know it worked

The Selectability Score (Part five) checks whether the buyer, or the machine, already knows you. Your own numbers go further, moment by moment:

• The share of new customers who name a chosen moment when asked what was happening.
• Searches for your name alongside each moment's words.
• The share of needs-based prompt runs that name you, for each moment.

Toolkit

1. Building the moment memory map. The process, start to finish. Run it once to set a baseline, then each quarter with the same prompts and questions.

Read it left to right, then down. Each moment is checked in both memories, and any gap loops back for a retest next quarter.

2. Moment memory map template. One row per chosen moment. Rate each memory strong, weak or absent, using the scale in the method.

Every template for this stage is in the free Selectability Playbooks at rocmode.com/playbooks.

What we don't know

• No peer-reviewed study I found shows that moment-based memory measures predict actual sales better than prompted awareness. The strongest link uses stated intention, from one B2B insurance survey.
• No published study tracks a brand causally from more coverage to more AI recommendation. The AI evidence is correlational, or uses invented brands, and both AI studies are preprints without independent replication.
• The one small-firm case here that was built from zero is my own, Cru Digital, and it's self-reported and from before AI assistants. VOCAL has a plan but no results yet.
• We don't know whether the two memories reinforce each other.

Whose evidence this is. Ahrefs sells SEO tools and an AI-visibility tracker. LinkedIn, which commissioned the insurance study, sells B2B advertising, and Ehrenberg-Bass is funded by corporate sponsors. The Snickers figures come from the agencies' own awards paper. Peer-reviewed sources: Hoyer and Brown; Romaniuk on Dirichlet norms. Chu and Hou, and Malthouse and colleagues, are preprints not yet peer reviewed, and Ludwichowska and colleagues is a conference paper.