The Brand Command System Mission 10 All missions

Mission 10 · Zero

Test so the answer means something

4 lessons Working asset: Range Card Needs a live account

Zeroing a rifle is firing a controlled group, reading exactly where it landed, making one adjustment, and firing again. Not spraying and hoping. That is what optimising an account actually is, and almost nobody does it that way. They change four things at once on a Tuesday night, the numbers move, and they have no idea why.

Lesson 1 of 4

Setting a test that can answer something

Most things people call tests cannot produce an answer. Two different creatives, in two different audiences, at two different budgets, judged after a day and a half. Whatever happens, you have learned nothing, and you will act on it anyway.

01 A test starts with a question

Before you build anything, write the question down. One sentence, in plain words, with an answer you could actually act on.

Not a question

"Let's see if this works."

There is no outcome of that which tells you what to do next.

A question

"Does the before and after beat the owner talking to camera, with the same audience and the same budget?"

Either answer tells you what to make more of next week.

02 One variable

If you want a clean answer, only one thing can be different. Same audience, same budget, same optimisation event, same offer. Change the creative only, or change the audience only.

And one honest caveat

You will not always be able to afford a clean test, and that is fine. Just be honest with yourself about what the result can and cannot tell you. A messy test is still worth running when the alternative is nothing. It is only dangerous when you treat a messy result as a clean one.

03 Test where it cannot do damage

New things go in the testing campaign. Never into a campaign that is working.

Separate campaign, separate budget, isolated. That is the whole reason Creative Testing exists as its own thing in Mission 04.

04 Fund it properly or do not run it

A test needs enough budget to produce enough results to be read. Work it out before you build: daily budget, divided by your expected cost per result, times the number of days. If that arithmetic produces two or three results in total, you do not have a test, you have a coin flip.

For a new audience, a small daily amount per ad set over three to seven days is the usual starting shape. Below that, widen the gap between what you are testing or test fewer things at once.

05 Test big differences, not small ones

At your budget this matters

You cannot detect a small difference on a small budget. Two headlines that are nearly the same, a slightly different colour, a tweaked first line: at the volume a first client runs, those differences disappear into noise and you will confidently read a winner that was random.

Test things that are genuinely different. A different format, a different angle, a different offer, a completely different audience. Big swings produce readable answers.

06 Decide the verdict before you launch

Write down, in advance, what result would make this a winner and what result would make it dead. Then you are reading a number against a line you already drew, instead of deciding in the moment whether you feel good about it.

THE QUESTION:
WHAT I AM CHANGING (one thing):
WHAT IS STAYING THE SAME:
WINNER IF:
DEAD IF:
I WILL JUDGE IT ON: [date]

07 Your task

Lesson 10 / 5 complete

Lesson 2 of 4

Enough data to decide

The two ways to get this wrong are deciding too early, which is nearly everyone, and never deciding at all, which is the people who overcorrected after reading about deciding too early.

01 Count results, not days

Days tell you how long something has been running. Results tell you whether you can read it. A week at a tiny budget and two days at a real one are not the same amount of information, even though one is three times longer.

Rough shape to work with: under three results, you do not have an opinion. Three or more and you can start to form one. And give it three to seven days regardless, because the platform needs time to settle before its numbers mean anything.

Both conditions. Not one or the other.

02 Colour it in advance

Same rules you set in Mission 05. Decide the lines once, then you are reading rather than judging.

Cost per resultMeans
Green, comfortably under targetScale it, by the rules in Lesson 3
Yellow, between target and the kill barLeave it alone and watch it
Red, at or over the kill barAct

03 What dead actually looks like

People say something "did not work" when it has spent about half of one result's worth of budget. That is not a verdict, that is impatience.

A fair definition of dead: it has spent several times your target cost per result and produced nothing. At that point the thing has had its chance.

Anything that has spent less than one result's worth has not been tested, it has been glanced at.

04 What you cannot judge it on

05 Zero is different

Zero is never noise

Small numbers are noisy. Zero is a signal, every time. Zero delivery, zero results after real spend, zero leads arriving when the platform says they were submitted. Check zeros the same day, because zero usually means something is broken rather than underperforming.

06 Your task

Lesson 20 / 4 complete

Lesson 3 of 4

Kill, keep, scale

Three decisions, and most accounts die from doing the third one badly or the first one too fast. Write the rules down once and then follow them, because the entire point of rules is that they work when you are tired and the client is messaging.

01 Kill

Killing is not failure. It is the point of testing. If nothing ever gets killed you are not testing, you are accumulating.

02 Keep

The one people skip. Not everything has to be a winner to earn its place.

Why keeping the mediocre ones matters

An ad that is not the star may be the only thing in the account speaking to one stage of the awareness ladder. Kill it and you have a gap, and spend concentrates further into the one or two ads that are already carrying everything. That is the fragile account from Mission 06, and it falls over the day the star tires.

Keep it if it is producing at an acceptable cost and covering something nothing else covers. Kill it if it is producing nothing or if something else already does its job better.

03 Scale, with actual numbers

Read these as results per ad set. Results means purchases for a store and qualified leads for a service business.

SituationMove
First week, two or three results, cost on targetUp by ten to twenty five a day
Three or more results, holding above targetUp ten to twenty five every three to four days
Fewer than three results, below targetDown twenty to fifty percent
Performance drops right after a scaleBack to the previous budget
No results at all for a week after scalingCut it
Already above two hundred a day and consistentUp fifty to a hundred every two to three days
RetargetingSlower than everything else. The audience is finite

Scale on consistency, not on a good day. One excellent Tuesday is not a trend. Several days holding is.

04 Up or sideways

Vertical

More budget into the same audience. Simple, and at some point it stops working because you have reached everyone worth reaching at that price.

Horizontal

New audiences at small budgets alongside. Use this when going up has broken the cost. Five ad sets at twenty each that work beat one at a hundred that does not.

05 The weekly migration

Winners do not stay in the testing campaign. Every week: proven creatives move out into the campaigns that carry real budget, tired ones come out of those, and five new ideas go into testing. That rotation is what stops an account ever having an emergency.

06 Change one thing, on one day

The rule that saves accounts

Budget changes happen on your review day. Not on a Tuesday night because you looked at your phone. Every change disturbs delivery, so an account that gets touched daily never has a stable period to be judged on, which makes you touch it again.

And change one thing at a time. If you move the budget, swap the creative and change the audience together, you will never know which one did it.

07 Your task

Lesson 30 / 5 complete

Lesson 4 of 4

The decision log

The least exciting thing in this entire course and one of the most valuable. You will not remember what you changed. You will be certain you do, and you will be wrong, and three weeks from now something will shift and you will have no idea what caused it.

01 What it is

One line every time you change anything in an account. Takes about twenty seconds. That is the whole discipline.

DATE | ACCOUNT | LEVEL | WHAT I CHANGED | WHY | WHAT I EXPECTED | WHAT HAPPENED

The last two columns are the ones that make you better. Writing down what you expected, before you know, is how you find out whether your instincts are any good. Most people have never once checked.

02 Four things it gives you

1

It explains the present

Performance shifts and you can see exactly what you did eleven days ago. Without the log you are guessing about your own work.

2

It writes your report

Your monthly client update is largely this log, tidied. Mission 11 depends on it.

3

It stops you going in circles

You will try the same failed thing twice. Everyone does. The log is what catches it the second time.

4

It turns into your own playbook

After a few months in one market, patterns appear that are specific to you and your clients. That is knowledge nobody can sell you and it exists only if you wrote things down.

03 Log the nothing too

Deciding not to change something is a decision. Write it down with the reason. When the client asks what you did this week, "I looked at it properly and deliberately left it alone because it is still inside the learning window" is a far better answer than inventing busywork, and it is only available to you if you recorded it.

04 Review it monthly

Once a month, read your own log back. You are looking for three things.

05 Your task

Lesson 40 / 4 complete

Mission 10 complete

You can design a test that answers something, wait long enough to read it, make the call by rules instead of by mood, and prove later what you did and why. Mission 11 is keeping the client, which is where the money actually lives.

Working assets

Range Card and decision log

One card per test, filled in before you launch it. One log line per change, forever. Together they are the difference between running an account and reacting to one.

01 Range Card

Filled in before the test is built. If you cannot complete it, the test is not designed yet.

RANGE CARD
Client:                          Test number:
Date launched:                   Judge on:

THE QUESTION
(one sentence, both answers must be actionable)

THE ONE VARIABLE

HELD CONSTANT
Audience:              Budget:
Optimisation event:    Offer:
Placement:             Creative format:

THE MATHS
Daily budget:                    Days running:
Expected cost per result:
Expected results across the test:
Is that enough to read?   [ ] yes  [ ] no

THE LINES, DRAWN NOW
Winner if:
Dead if:
Inconclusive if:

THE RESULT
Spend:          Results:         Cost per result:
Verdict:   [ ] winner   [ ] dead   [ ] inconclusive
What I actually learned:
What I am doing because of it:

02 Decision log

DATE | ACCOUNT | LEVEL (campaign/ad set/ad) | WHAT I CHANGED | WHY |
WHAT I EXPECTED | WHAT HAPPENED | REVIEWED ON

Twenty seconds per row. Fill it the moment you make the change, not at the end of the week.

03 My rules, decided once

Print this and keep it next to you. It exists for the night when something looks bad and you want to do something about it.

MY RULES

Target cost per result:
Kill bar:
Review day:                    Changes happen ONLY on this day.

DEAD means: spent [    ] with [    ] results
ENOUGH TO JUDGE means: [    ] results AND at least [    ] days

KILL
- Real spend, no results
- Working but far more expensive than my alternatives
- Nothing after a second fair stretch

KEEP
- Producing at an acceptable cost
- Covering an awareness stage nothing else covers

SCALE
First week, 2-3 results at target     -> up 10 to 25 a day
3+ results holding above target       -> up 10 to 25 every 3-4 days
Under 3 results, below target         -> down 20 to 50 percent
Drops right after a scale             -> revert
No results a week after scaling       -> cut
Above 200 a day and consistent        -> up 50 to 100 every 2-3 days
Retargeting                           -> slower than everything else

ALWAYS
- One change at a time
- Log it the moment I make it
- Zero gets checked the same day

04 Weekly review

Review day0 / 8 complete

Field note

The best media buyers are not the ones with the best instincts. They are the ones whose instincts have been checked against a written record often enough that they know which of them to trust. That is all the log is doing.

The Brand Command System · Mission 10 Zero · Recon. Deploy. Payload.