Your words from this morning: you need "volume right now," and you need "diversity." The budget went from $85K to $30K when the CAC target moved, and the path back up runs through creative, not media buying. So here is the one-month test, structured: 14 spec ads built before anything was signed, from your live ad account, your own verified studies, and 60-plus verbatim Reddit quotes. No month-to-month commitment. If the team says go on Thursday, kickoff starts this week.
$30K/mo today, was $85K14 spec ads, five lanes openedCAC target in flux: $185 to $100Meta only. USD.
Format diversity is fine. Persona diversity is the gap.
This is real data, so we label it: a gap analysis of your live Meta ads, run before the call. The format spread is actually decent, testimonials, bullet-point statics, product shots. The persona spread is not. Meta effectively buckets every ad you run as one audience: new parents. The slate below attacks exactly that.
Your live ads, as Meta reads themThe 14-spec slate
Persona concentration · your live ads
New parents
~90%+effectively all
The 3am symptom searcher
~0gap
The evidence-demanding parent
~0gap
The gut-test-report holder
~0gap
The formula-feeding optimizer
~0gap
The value-conscious subscriber
~0gap
The five lanes the 14 specs open
The evidence-demanding parent (HCP-influenced)
5 specs02 · 05 · 07 · 11 · 12
The 3am symptom searcher
3 specs06 · 10 · 14
The gut-test-report holder
1 spec03
The formula-feeding optimizer
1 spec08
The value-conscious subscriber
1 spec09
19 active ads
Live library, Sep 2
A real format spread: testimonials, bullet-point statics, product shots. Formats are not the problem.
1 persona bucket
Meta's read
Effectively every live ad categorized as "new parents." One audience, nineteen ways.
0 receipts ads
The untouched moat
No 9-in-10 stat creative, no published-RCT creative. The category's deepest proof, sitting unused.
12 sub-personas
The slate
Fourteen specs, twelve distinct sub-personas of "new parent," weighted toward the proof angles.
"The way that Meta is categorizing all of our personas as new parents is mind-blowing to me."Kira, on the call. That sentence is this slate's brief. When every ad speaks to the same bucket, the algorithm has one pocket to fish in. Five new lanes give it five, each one speaking one-to-one to a parent your current account never addresses, plus three brand-and-proof slots (01 · 04 · 13) that work across all of them.
Every lane above maps to a spec on the work tab. None of them recycles an angle you already run.
Kira's own words from September 2, and how the test month is built around each one.
"We were up to 85 for a while… they cut budgets to like 30."
Kira · on the call
Budget gets rebuilt on proof, not promises. The test month exists to earn the scale-back-up: creative strong enough that the CAC math makes the argument for us, so the next budget conversation starts from results.
"They're all new parents because our product is very specific for infants and toddlers."
Kira · on the call
One persona bucket is why accounts stall. Five new lanes, each speaking one-to-one, so the algorithm can find pockets the current account never addresses: the 3am searcher, the evidence demander, the gut-test holder, the formula optimizer, the value shopper.
"We're like a marketing team of two… it's an educational process."
Kira · on the call
Turnkey volume without headcount. Concepts arrive written, graded, and ready; Kira reviews instead of produces. Bi-weekly calls keep the team inside the process without adding to it.
"She's like anti obvious AI ads… I don't want fake babies. It's all about authenticity."
Kira · on the call
Standing rule for this account: zero faces, zero babies in AI renders. Baby presence is objects only: bottles, sippy cups, post-its. Every line of copy traces to real customer language, not model hallucination. All 14 specs on the work page already obey this.
Worth saying out loud: nothing here requires a brief from your side. Week 1 starts in your ad account, not your inbox. Brian's positioning docs fold straight in, persona-first, and the process section below covers exactly what we need (account access) and what we do not (briefs, headcount, extra meetings beyond the bi-weekly).
Proof
What clients say when the volume lands.
Real notes from the roster, unedited. The same engine, the same cadence, the same team that built your 14.
Brodo
"We've had a fantastic year, and you've been a big part of that success. Your team's work has consistently exceeded expectations."
AndrewCEO, Brodo
OpenStore
"We've been very happy with the quality we've received, the improvement you've shown throughout our partnership, and the excellent attentiveness and customer service you've given us."
Billy BjorkCreative Operations & Strategy Lead, OpenStore
The Cumin Club
"We had a strong Feb and March, excited to keep the momentum going. Let me know if your team can support more volume; looks like that's the only way to scale further."
Ragoth BalaFounder & CEO, The Cumin Club
Process
No briefs. That is the process.
The question from your team's pre-meeting email: do we need to write a brief? No. The system runs on your ad account, your customers' own words, and the positioning work Brian is already doing. Here is the shape of the month.
Week 1
Plug in and tag everything
We plug into the ad account and tag every ad you have ever run: persona, angle, format, emotion. That map is what makes the gaps visible and keeps every new concept a genuine test instead of a rerun.
From your side: account access. That is the whole ask.
Then
Gap analysis + social listening
Reddit threads, reviews, comment sections: real customer language becomes the angle bank. Brian's positioning docs get folded in, persona-first, so the strategy work you are already paying for shows up in the ads.
Why it matters: copy traces to quotes, not guesses. The CEO's Reddit instinct, systematized.
Ongoing
60 / 20 / 20
Every batch splits three ways: 60% iterate winners, 20% reskin winners, 20% test net-new personas. Losers get killed by data, winners get built into families, and the net-new slice is how the five lanes keep widening.
Cadence: weekly concept delivery, bi-weekly calls, everything tagged and queryable.
Why now
September is the testing window
Q4 CPMs punish late testers. Testing in September means Q4 spends into proven winners instead of paying holiday prices to learn. Kickoff can start this week once the team says go.
The frame: one-month test, no ongoing commitment, results before the Q4 budget talk.
Pricing
Three tiers. Your spend picks one.
SelfMade pricing, applied to Persephone, in USD. Slide to your monthly ad spend. The tier snaps in steps: Starter under $40K, Standard from $40K, Scale from $80K. At today's $30K, Starter is where the test starts. Month to month, no commitment: the one-month test is the deal on the table.
Every tier includes the full engine: gap-analysis briefs, 9-rubric copy refinement, Agent-10 grading on every render before delivery, the weekly cadence, quarterly Adlib re-runs, and time-series plus social listening bundled. No extra cost.
What counts as a concept: 1 concept is 1 static ad. A video counts as 6 statics. Mix them however the account needs: Tier 1's 8 concepts a week can be 8 statics, or 1 video plus 2 statics.
Your monthly ad spendSlider at $30K / mo
$30K/ month
$6,000 / mo creative
$20K$40K$80K$125K
Recommended for your $30K
Tier 1 · Starter
$6,000/ mo
Up to $40K/mo ad spend
Where the one-month test lives.
8 unique concepts a week
Recommended for your $30K
Tier 2 · Standard
$9,000/ mo
$40K to $80K/mo ad spend
The road back to $85K runs through here.
14 unique concepts a week
Recommended for your $30K
Tier 3 · Scale
$15,000/ mo
$80K to $125K/mo ad spend
For when the budget comes all the way back.
30 unique concepts a week
Add-on · landing pages
+$3,000/ mo
Persona-tuned landing pages
Persona-tuned LPs spun up against your winning ad concepts. Recent comp: roughly 30% CAC reduction from adding the layer alone.
Kira's words: "definitely not in the budget for this test run." Off by default; the number is here so it exists when it is.
Add landing pages
Add-on · media buying
+$2,100/ mo (7%)
We run the Meta buy
Done-for-you Meta media buying. $2,000 a month flat under $30K in monthly ad spend, 7% of spend at $30K and above.
Not part of this test: the buying conversation is separate. Off by default; the price is on the record for later.
Add media buying
Your proposal total
$6,000/ month
TIER 1 · STARTER · 8 CONCEPTS / WEEK
Tier deliverables are fixed: 8, 14, or 30 unique concepts a week. 1 concept is 1 static; a video counts as 6. Final start date locks when the team says go.
↑The road back to $85K, and past it: the creative fee only steps up after the spend does, so scaling the budget back is what moves the tier, not the other way around. Above $125K/mo in ad spend, the contract has a clear upward path against the same per-creative economics.
Thursday, 9am PST. Already on the calendar.
The team meeting is tomorrow, Thursday September 3 at 9am PST, already booked. This page is the pre-read: the persona data, the 14 specs, what we heard, the pricing. Bring the detailed questions Thursday, we saved time for them.
And if the team says go: kickoff can start this week. September is the testing window; Q4 CPMs punish late testers.
.
