ArticleIssue #186 | Allocating The One Resource That Matters
I hope you’re all doing well, enjoying the start of Fall and excited for 2026’s home stretch. Conference season is officially back in session for me – I spent most of last week at HeroConf in San Diego. While the event itself was OK, the people were fantastic – in fact, one of the conversations I had there was the genesis of this very issue.
In a conversation with an old friend, I made the comment that there is only 1 input your marketing team/agency/organization can’t buy more of. It’s not budget, headcount, media, ideas, insights, SaaS or production capacity (in fact, with AI, production costs have fallen so far that they barely function as a constraint at all). Doing more of any of those things is – for almost every organization – relatively trivial.
In fact, I’ve come to the conclusion that the ONLY input that doesn’t scale is the judgment of the handful of people who are actually good at this. You have what you have, it costs what it costs, and nobody is selling more of it. Taste, nuance, skill, tact, discernment – whatever you want to term it – is a true competitive advantage.
If you accept that, it leads to a VERY interesting operating analysis: if that’s true, then great marketing is genuinely an allocation question: specifically, what is the highest & best use of that resource?
If you’ve been around for a while, you might remember that I ran that very audit on my own team and published the result back in February (Issue #156). The conclusion was both eye-opening and organizating-changing: ~65% of the recurring decisions on our weekly calendar were determinable. Not easy or simple or elementary. Entirely determinable. And by “determinable”, I mean if 5 competent analysts examined the same data with the same business context, they would land in substantially the same place.
Put in the parlance I’ve used throughout this newsletters run: the are math questions.
Math questions have answers.
The examples go on and on:
- What a conversion is worth (i.e. what you cap your bids at) – Issue #183, where the correct average ROAS target is 1 ÷ (elasticity × contribution margin) and the marginal number that controls is 1 ÷ margin (see: Issue #183).
- The spend where the next dollar stops paying for itself: marginal CAC vs. unit economics (see: Issue #162)
- How much creative a given budget can carry before individual ads starve (see Issue #158).
- How long you have to wait before a difference between 2 things is real = ~4 ÷ √N (see: Issue #182).
- How to distribute a fixed budget across a month = evenly, almost always (see Issue #179).
- Whether your measurement window covers the decision it’s measuring (Issue #184).
- How much more budget you should allocate to a given product/service/campaign/property (Issue #183).
The list goes on and on. Every one of these questions has an answer.
The problem is marketers have convinced themselves that these questions “depend” on stuff – feelings, vibes, expertise, intuition, precedent, experience, testing. That’s bullshit.
The reality is that most marketers simply inherited a “refresh ads every month” scope item or a 4x ROAS target or a uniform advertising budget per community or a nurture series, and have never bothered to check whether it makes sense (hint: it probably doesn’t). That’s a very expensive mistake.
But…Not Everything That Isn’t Math = Art
Earlier this year, when I wrote “The Art & Science of Media Buying”, I drew a binary split: if it’s not math, it’s art. That…may have been incomplete, so I’ll amend my previous statement to this: There’s math, there’s art, and there’s better starting points.
Some examples:
- No formula produces an account structure
- No derivation produces a landing page
- There is no mathematical path from unit economics to naming conventions
But, just because there’s no absolutely optimal (or correct) answer doesn’t mean that all possible answers are equally right (or equally wrong). To the contrary – there are better answers to each of those questions, which can be arrived at via accumulated evidence (vs. math).
An account whose architecture mirrors the business (e.g., divisions by product line, margin tier, audience value, funnel stage) will out-perform one that accreted as new campaigns got heaved in next to the old ones. A page that answers the 10 questions a visitor is actually asking (Issues #167 and #168) will beat one that opens with a company history.
None of that is taste. It’s just better starting points.
Those are (at least, to be) fundamentally different from “best practices” (which are almost always bullshit). A best practice is a convention you inherited without its rationale. You can’t say what problem it solved, for whom, under what conditions, or what evidence would cause it to go away. A better starting point is the same convention with all 4 still intact. That makes one a commandment you obey, and the other a default you intentionally adopt UNTIL the day comes when the conditions change. The content might be identical, but the loyalty + understanding are not.
Surprisingly, that makes a VERY big difference.
Table Stakes vs. Advantages
I genuinely believe that these two items – the math + the better starting points – are likely to become commodities in the next 18 months, simply because (1) the answer exists before you arrive (i.e. you’re retrieving it, not producing it) and (2) the cost of getting that answer has become stupidly cheap thanks to AI (before, you’d need an entire data team to help you figure out a beta value; now Claude can do it in ~82 seconds – yes, I know, oddly precise).
That’s the harsh reality: your competitor has the same modeling capability you do. The major ad platforms automated a good share of the first category outright, then gave everyone the ability to use it. The second has been documented, shared and podcasted ad nauseum for much of the last ~5 years, which means anyone with working ears and a pulse knows about it.
My best-practices argument came down to one line: as adoption increases, surplus value decreases. A convention/setting/whatever that everybody has already adopted therefore returns nothing to the next person who adopts it.
Net/net = being excellent here buys parity and little-to-nothing else.
To be fair, parity isn’t nothing. Losing it is expensive, both in terms of the time required to gain it back AND the budget you’ll light on fire doing so. But, it’s still table stakes, paid in a currency your competitors also hold. If you want a Sam-ism: adequacy is free, which is exactly what makes it worthless.
If we go back to one of this year’s most popular issues (seriously, it’s ~2nd most viewed one~): Creative Is Math Wearing a Costume (Issue #164), you’ll see that very idea play out in a narrow, specific way. We know that AI has caused production costs to collapse, and the binding constraint has moved from how many ads you can make to what share of them win. We all know hit rate = a function of the strategic quality of what you chose to test, not the quantity you managed to load into Meta before it broke for the 19th time in 24 hours. Basically: cheap retrieval never raises your ceiling, but it DOES lower everyone’s floor on the same schedule.
When you look at every decision you have to make, the honest reality is that the residual – the stuff reserved for art – is miniscule:
- Which emotional real estate your category/industry has left unclaimed
- What offer/offer type your prospective customer wants, but can’t find
- What that random spike in the data a few weeks ago was telling you
- Which audience a reframed offer could unlock that your positioning currently excludes
- Whether a competitor’s move is a threat or a tell
- What the buyer feels about the purchase, vs what the survey said they feel.
For any one of these questions, 5 competent analysts produce 5 different answers; the distance of those answers and the worst is worth more than everything above it combined.
Taste, discernment and judgment only pay in this category. Everywhere else? They’re decorative.
Unfortunately, the reason the misallocation survives is because arguing about account structure feels like strategic work while you’re doing it – it just isn’t. The same is true for the number of ads to run or the target to set or the geo to target – it “feels” fun. Though, take it from me, it’s not. No one enjoys arguing about account structure or the number of ads to run or whether a given ad should be left alone to run or culled.
Intentions & Deadlines
You might read this and think, “We already have this solved.” Reality check: I’ve looked at hundreds and hundreds of ad accounts. The probability says you’re wrong.
When we were having this conversation, my friend even mentioned that their team has built an entire system to counteract this very thing: 4x annual off-sites at a nice place, with drinks included (not sure why that detail was relevant, but I feel compelled to keep it here for posterity), super-fancy-premium tech tools AND some AI agents that go through and (allegedly) flag the “math” stuff so it doesn’t suck up time.
But that gets the question wrong.
The question isn’t whether strategic work is on the calendar. It’s what happens to it on Thursday when pacing looks wrong, a CEO asks why CPL increased 11% WoW & somebody needs an answer by 4 pm.
The reality is this: retrievable work has deadlines. Artistic work has intentions. Deadlines win every time. And not only do they win, but they do so by consuming the exact people whose judgment was the only reason the artistic work would have been any good.
Which means that fancy offsite is the overcompensation (i.e., an admission that day-to-day operations of the marketing org has no room in it for thinking so thinking gets scheduled 4x a year, off site, by exception). Not great.
What To Do About It
I don’t have a math answer to this question, BUT I do have some better starting points.
Step #1: Take the standing questions your team deals with every week. Categorize each: computable, known or open. It will not be fun. A large share of what a senior team experiences as strategic debate turns out to be math solved in the least efficient manner OR a convention debate that’s been resolved in your industry/vertical for a half a decade, but no one has bothered to look.
Step #2: Resolve the first 2 once, in a written AND shared document, with the rationale and the conditions that would reopen them. That’s the difference between a default and a dogma; it’s also what stops the same argument from restarting every time somebody joins who has opened up SearchEngine Journal recently.
Step #3: Protect what you recovered. Believe it or not, recovered time does not flow toward judgment on its own (ask me how I found that out). It flows toward whatever the review meeting asks about. If the Tuesday meeting focuses on pacing, then pacing is what gets prepared, which means your best people will spend Monday pm on it. Change the question the meeting asks and you change where the judgment goes. Incentives are weird + miraculous things.
Contrary to popular (and my own prior) beliefs: the highest and best use of processes is not to limit what talented people can do; it’s to clear out the cognitive load that stops your smart people from doing the smart stuff you hired them to do.
So, solve what can be solved. Adopt what’s already known. Doing that for those 2 categories is well on its way to being free for everyone. That leaves the art – the tastemaking and discernment and judgment – as the last bastion of true creative differentiation. Your job – whether you’re the CEO, CMO, Agency owner or just the person that got stuck with marketing – is to ensure that your smart people spend as much of their time on the art as possible.
