Strategic Approach

Seven calls where the obvious move is the wrong one.

Knowing the frameworks is table stakes, since anyone can watch the same videos. What decides an account is the call you make when the numbers point the wrong way. So every exhibit here ends in a decision, argued on the ads in this Library.

Simulated numbers on real creative. The ads are the ones in this Library, the figures are invented and sized to DTC supplement economics, and the reasoning is what I would apply to a live account.

The ad you'd cut is the one paying for the rest.

Argued on PetHonesty Dog-to-Dog · NeuroGum Office Authority

Both lines rise together, so the ad with the worst multiple is also the one still reaching new people.

The obvious call

Dog-to-Dog is running 0.7× after three weeks. Office Authority is at 6.4×. Cut the loser, move its budget to the winner.

The call I'd make

Keep Dog-to-Dog on and stop funding Office Authority. Dog-to-Dog runs at frequency 1.1, so it is still finding people nobody has reached. Office Authority sits at 3.6 and is closing buyers that Dog-to-Dog opened.

Why. Delivery is a sequence and the dashboard credits the last click, so reported ROAS climbs toward most-aware while frequency climbs with it. Judge at the ad set.

Your best ad will not take more budget.

Argued on NeuroGum Caffeine and Meth

The multiple is identical. The spend each one can still absorb is not.

The obvious call

Caffeine and Meth is the account's best performer at 4.1×. Double its budget and take the revenue.

The call I'd make

Check frequency before touching the budget. At 1.2 I raise about 20% every second day. At 3.4 I leave it alone and spend the money on cold reach, because the multiple is measuring harvest.

Why. Total profit tracks ROAS multiplied by the spend a concept can hold. A high multiple on a saturated pocket has no ceiling left to buy.

The same 3× is a win and a disaster.

The creative performed identically in both rows. Only the margin moved.

The obvious call

The account is at 3.0× ROAS. That clears the 2.5× the client asked for, so the creative is working and we scale.

The call I'd make

I ask for contribution margin before I read the number. At 40% margin, 3.0× is profit. At 24% margin, break-even is 4.17× and that same 3.0× is losing money on every order.

Why. Break-even ROAS is one divided by contribution margin. A ROAS target borrowed from another account is a number with no meaning attached.

Nine variations buy one ticket.

Argued on PetHonesty The Walk That Keeps Getting Shorter

Volume on an unproven idea multiplies a loss. Volume belongs underneath something already proven.

The obvious call

Walk-Shorter is winning. Cut three alternate openings and three thumbnails from it, ship nine variants, and give the winner nine chances to be picked.

The call I'd make

Ship three genuinely separate concepts instead. Same insight, rebuilt as a benefit lead, a problem lead, and a story lead, which the system reads as three different ads.

Why. Near-identical creatives collapse into one container and compete as a single entry, so nine variations buy roughly one shot at delivery.

A 7× that is actually a 1.9×.

The obvious call

The catalogue campaign returns 7.2×, far above everything else in the account. Shift budget into it and scale the thing that works.

The call I'd make

I re-read it on 7-day click with view-through excluded, and check it against MER. Catalogue ads intercept people already on their way to buying, so most of that 7.2× was revenue the brand had already earned.

Why. Feeding a harvester raises reported ROAS while blended MER stays flat, which is the signature of paying twice for the same customer.

The angle dies before it costs anything.

Argued on Ancient Nutrition · the Wave 2 expansion angles

Any one failure kills the angle. The gate that kills most often is the elevator test.

The obvious call

Eight new angles are queued. Produce all eight, test them, and let the data decide which ones work.

The call I'd make

Run all eight through four gates first, and expect three to survive. A produced concept tested for five days costs roughly $500 in spend plus production plus a week, and the answer was available for free.

Why. Testing confirms which survivor performs. Curation is what produces a hit rate, and hit rate is the number that actually measures a strategist.

Checking on day two is how you buy a false winner.

The obvious call

The test is live. Check it every morning and cut the losers early so budget stops burning on ads that are clearly behind.

The call I'd make

I write the window and the thresholds before launch and do not look at performance until the window closes. Daily checks get delivery and spend only.

Why. Stopping the moment a result looks favourable converts a test into a random draw, and the early leader reverses often enough that the habit is expensive.

Four reads I use most, and what each one changes.

Single metrics rarely diagnose anything. Every read below is a pair.

Low hook, strong hold

Means

The opening fails and everyone who survives it is the right person.

Keep the body untouched, rewrite the first three seconds. Highest-leverage fix available.

Strong hook, low hold

Means

You bought attention from the wrong people. Suspect clickbait.

Fix pacing and structure, or pull the hook back toward the actual promise.

High CTR, low CVR

Means

The ad over-promises or stays vague, so the click is unqualified.

Tighten the ad's specificity so it self-qualifies. Leave the lander alone for now.

Ugly soft metrics, real profit

Means

Heavy self-qualification. Few people stop, and the ones who do are buyers.

Nothing. Never turn off a profitable ad because its engagement looks unimpressive.

One account, diagnosed in order.

CPA is $86 against a $52 target and the client wants new creative. The procedure runs top down and stops at the first number that is off, which means most of these checks never run.

01
Is the primary number actually off?
Continue
CPA $86target $5265% over
Off target. Keep going.
02
Are the ad metrics off, measured against this account's own average?
Continue
Hook 27% vs 24% avgHold 8.1% vs 7.4% avgOutbound CTR 2.9% vs 2.6% avg
Every one is above the account's own average. The ad is doing its job, so the break is after the click.
03
Are the funnel ratios off?
Stop here
Link click to landing page view: 71%the bar is 80% or betteradd to cart 9.4%, healthy
Stop here. Roughly 29% of paid clicks never reach the page, which is a load-speed or redirect fault. No creative fixes that, and the real number comes from Shopify or Clarity rather than Ads Manager.
04
Is AOV or the offer the constraint?
Never ran
Never ran.
05
Did the ad buy the wrong people?
Never ran
Never ran.

The conclusion. The brief that came in asked for new creative. The diagnosis says the creative is the strongest part of this account, and a third of the traffic it bought is being lost between the click and the page.

The same reading, on three ads from this Library.

Each of these commits to a prediction before the spend exists, and names the number that would prove it wrong. A prediction that cannot fail is decoration.

PetHonesty

The Walk That Keeps Getting Shorter

Problem aware

Built to

Opens on eight months of walking-app data and withholds the product until the diagnosis has landed. It is built to meet an owner who already noticed the decline and filed it under aging.

So I'd expect

Frequency near 1, because a problem-led open retrieves into colder pockets
ROAS below the product-aware ads in the same account, and that is the design working
Hook rate carrying the asset, since the data screenshot is the entire stop

What changes my mind

Frequency past 2 while ROAS stays flat means the problem-lead is landing in the same pocket as everything else, so I rebuild it against a different belief instead of re-hooking it. A collapsed hook rate with the hold intact means only the screenshot failed, and the first three seconds get rewritten while the body stays untouched.

NeuroGum

Caffeine and Meth Hit the Same Pathway

Problem aware

Built to

Leads on a receptor claim that reframes the category before the product is named, so the pattern interrupt and the mechanism arrive in the same line.

So I'd expect

A hook rate well above account average, because the claim is genuinely arresting
Hold rate as the number that decides it, since a claim this sharp stops people with no intention of buying gum
CTR that needs reading against CPM, never alone

What changes my mind

A high hook with a collapsing hold means the claim is pulling the wrong crowd, and the repair is the second beat rather than the first. High CTR against low CVR means the reframe over-promised, so I tighten the ad's specificity until it self-qualifies and I leave the lander alone.

Ancient Nutrition

The Choosing Self

Product aware

Built to

Opens on a scheduling detail the buyer recognises instantly and sells an identity rather than a mechanism. It sits at the warm end of the rotation on purpose.

So I'd expect

The best reported ROAS of the three, because it closes rather than opens
Frequency climbing faster than anything else in the account
A spend ceiling that arrives early, since the pool it speaks to is small

What changes my mind

If it reports 5× or better I still do not raise its budget, because that multiple is measuring harvest. Frequency is the number I act on: past 3 it is re-hitting a pool the openers stopped refilling, and the correct response is more cold problem-led work rather than more of this.

The week this runs on.

Different metrics move at different speeds, so reviewing them on one schedule guarantees over-reacting to some and under-reacting to others. The smaller the slice and the shorter the window, the less it is allowed to decide.

Launching Wednesday or Thursday is deliberate, since Monday and Tuesday are the weakest days in most accounts and the test should run through a weekend before Monday's read.

Two questions that come up, answered short.

A client asks for 10 ad concepts. What do you ask first?

Ten is somebody's proposed solution, so the first job is finding the problem it was meant to solve. Three things decide the brief, and none of them is a creative question.

What is driving the request: out of winners, one winner fatiguing, or spend stalled at a ceiling?
Contribution margin, AOV, and new-customer CAC, because without them I cannot tell you what working means
Where spend currently sits across awareness stages, and what account frequency is, which decides whether you need reach or conversion
What died recently, and what you concluded about why, since a wrong conclusion is more expensive than a dead test
What can actually be produced inside the timeline, because ten concepts you cannot shoot is zero concepts

What does your testing framework look like?

That every decision is pre-committed instead of improvised. Each batch varies one thing at the concept layer, thresholds and duration are written before launch, one primary metric is nominated in advance, and kill, keep, and scale rules exist before any data does. An ad set holds one concept with three to five variations inside it, so Meta searches for the best expression of the concept instead of the concept riding on a single execution. Every test lands in a log with its hypothesis, its pre-committed threshold, its result, and whether the read was statistically real or directional only.

The sentence written before every batch

We believepersonadoes not buy because they believecurrent beliefIf weroute to the new beliefthey willactionWe know insidewindowifone primary metricclearsthreshold set before launch.

The pipeline that produces the creative these calls are made about is in the Creative Approach, and the ads themselves are in the Library.