Lead Engine OS
Ad Management Playbook
01
Small hinges swing big doors
Test in this order. Every time. No exceptions.
1
Niche
Biggest driver of performance. Get the niche right before testing creative. Everything else is downstream of this.
→
2
Creative angle
Pain vs. gain. One variable changes between ads in the same test. Only after niche is confirmed.
→
3
Offer
Don't test your offer until you know what creative angle works. Only after a winning angle is confirmed.
02
When does a client transition from MVC to BAU?
There is no fixed date — it's a confidence call
Principle
The transition happens when you're confident you have something worth scaling — when the CP-Book looks good and you're ready to move it into volume. It might be a week, it might be a month. There is no formula.
In practice
When you launch, more budget goes into testing. As winners emerge, you shift budget toward what's working. The aim is always to spend more on what's working. Transition is the point where you're ready to formalize that shift into a CBO scaling campaign.
03
Creative testing — how granular?
Art, not science. Start big. Get granular over time.
Principle
Art, not science. Test based on feel and judgment — there is no rigid formula for how many variables to change. Your instinct is a data point too.
In practice
Start with bigger things, get more granular over time. Early tests: angle-level (pain vs. gain). Later tests: hook variations within a winning angle. Don't micro-test before you've found the macro winner.
04
Advertising is an uncertain game
You don't have to be right. You just have to be great at testing things.
At its core, every ad is an educated guess. The best creatives in the world are still guessing — they're just more systematic about it. There's an art to this and a science to it. The science is in your split tests and your data. The art is in the judgment calls between them. Accept both.
One thing that doesn't change: the cost of advertising goes up every year. More companies are spending online, demand for ad inventory increases, and prices follow. There are short-term exceptions — Covid pulled budgets back and made placements cheap; US elections flood Facebook with political spend and drive up CPMs temporarily; Black Friday and Christmas spike costs across the board. But the long-term trend is up.
That means the mindset of "we need to get the CPA cheaper every year" is broken. You're working against yourself. The right frame is volume and acceptable CPA. Clients need a certain number of customers per month to operate — that's the real goal. Set a volume target, agree on an acceptable cost per acquisition, and work within that. If the cost of ads is rising, the answer is pricing up, building lifetime value, and creating upsell paths — not chasing a lower CPA that the market won't support.
Why guess when you can test? You don't have to be right — you just have to be great at testing things.
05
Protect the blowouts
Knowing what to kill is the most important skill in this job
You keep accounts performing by protecting blowouts. A blowout is wasted spend you can never get back. Even if you never scale another ad, you can always increase spend on a winner to catch up — but a blowout is already on the books. That money is gone.
Miss a scale opportunity
You can catch up. Increase spend on the winner later. The opportunity cost is recoverable.
You can catch up. Increase spend on the winner later. The opportunity cost is recoverable.
Let a blowout run
You can't get that spend back. It's gone. This is why kill rules exist — not to be aggressive, but to protect the account.
You can't get that spend back. It's gone. This is why kill rules exist — not to be aggressive, but to protect the account.
How this works: Kill rules fire at specific checkpoints — Day 3, Day 7, Day 14, Day 30. Between checkpoints, log data and don't touch. The channel determines your spend threshold. Never kill on Day 1–2 data.
Winning the Week: Every Monday starts with a structured review before any changes are made. No mid-week changes without a flag reason. All decisions are logged. Full 7-day windows required before any kill/keep/scale call.
Monday — winning the week meeting
ReviewPull full funnel numbers per client per source. Where is the leak this week? Schedule rate → show rate → offer rate → close rate. Find the drop.
ReviewAd performance by ad. Apply kill/keep/scale rules. Every ad gets a call. No ads left in limbo.
PlanSet the work plan for the full week. What gets launched today? What creative needs prepping by Thursday? No bottlenecks mid-week.
LogUpdate the account log. What was tested, killed, or scaled? This is the most important deliverable of the week.
EscalateAnything outside authority: budget increase >30%, kill entire campaign, client churn risk. These go to Cam before action is taken.
Weekly rhythm
Monday
Winning the Week meeting + all kill/keep/scale decisions
Launch new tests based on last week's prep
Tuesday
Day 3 checkpoint for anything launched Friday/Saturday
Check new leads — UTM tags intact? GHL follow-up firing?
Wednesday
Mid-week budget check — minor adjustments only, no new launches
Flag anything trending badly for Monday — don't kill yet
Thursday
Creative prep day — all assets for next Monday's launch due by EOD
Day 7 checkpoint for anything launched last Thursday
Friday
Launch new tests (captures weekend data in the 7D window)
Log the week — what ran, what stopped, what's the read
Weekend
Monitor only — no changes. Screenshot anomalies for Monday.
Decision authority
MVC
Minimum Viable Campaign — ABO · Weeks 1–4 · General framework, client strategy may vary
Goal
Get one winning creative and one winning audience within 4 weeks. Find one thing that works — don't try to find everything.
Structure
1 Campaign · 1–2 Audiences · 2–4 Creatives per audience
Campaign
Ad Set 1
Audience 1
Ad Set 2
Audience 1
Ad Set 3
Audience 2
Ad Set 4
Audience 2
Ad 1
Creative 1
Ad 2
Creative 2
Ad 3
Creative 1
Ad 4
Creative 2
ABO — 1 ad per ad set
2 audiences × 2 creatives = 4 ad sets
Minimums
$30 per ad
Kill logic
Run minimum 48 hours before any changes · Follow client-specific kill rules to kill and reallocate budget
Budget
MVC budget is separate from retargeting. Retargeting runs simultaneously but is budgeted separately. Never let retargeting cannibalize cold audience testing data.
BAU
Business As Usual — 2 campaigns running simultaneously
Scale campaign
CBO · 70–90% of budget
Proven winners only. Never test here. Winners graduate from the test campaign into this one.
Test campaign
ABO always · 10–30% of budget
New creative testing only. 1 new test per week minimum. Winners graduate to scale. Losers get cut.
Scaling
Follow client-specific scaling rules
Testing
1 new creative test per week minimum. Always have something in the test campaign. Winners don't last forever — feed the pipeline constantly.
AI
Meta AI defaults — applies to both MVC and BAU
Global rule
Any AI feature on any platform is off by default unless the client strategy explicitly says to enable it. If it's not stated, it's off.
Ad set level
OffAdvantage+ Audience — manual targeting always
OffAdvantage+ Detailed Targeting — manual only
OnAdvantage+ Placements — on by default unless stated otherwise
Ad level
OffMulti-Advertiser Ads — shows your ad alongside competitors
OffAdvantage+ Creative — master toggle, turn this off first
OffImage enhancements — auto-crops, filters, brightness
OffText enhancements — rewrites your copy per audience
OffAdd music — adds background music to video ads
Off3D animation — adds motion to static images
OffRelevant comments — shows user comments on your ad
OffEnhance CTA — Meta changes your CTA text
MVC
Minimum Viable Campaign — Weeks 1–4
Goal
Get one winning ad and one winning keyword set within 4 weeks.
Structure
1 Keyword Set → 3 Ads. Start with highest-intent exact match keywords only.
Monitor
Check search terms every other day. Add negatives as junk terms appear.
Rule
No broad match in Week 1. Exact and phrase match only until you have search term data you trust.
Budget
MVC budget is separate from retargeting. Never let retargeting cannibalize cold search testing data.
BAU
Business As Usual — ongoing management
Scaling
Optimizing for a lower 60-day cost per lead or cost per booked call is the primary decision signal for Google. Follow client-specific scaling rules.
Testing
1 new ad test per week minimum. New headlines, descriptions, or landing pages. One variable at a time.
Search terms
Review search terms every other day. Add negatives weekly. This is ongoing — it never stops.
AI
Google AI defaults — applies to both MVC and BAU
Global rule
All AI features off by default unless stated in client strategy.
OffSmart bidding (tCPA / tROAS) — use manual CPC until you have enough conversion data
OffBroad match keywords — exact and phrase match only until stated otherwise
OffPerformance Max campaigns — do not launch unless explicitly in client strategy
OffAI Max — a separate feature from Performance Max. Expands keyword matching and rewrites ads using AI. Off by default.
OffAuto-applied recommendations — turn off in account settings. Google auto-applies changes without notice.
MVC → BAU
Transition triggers — all must be met · Cam sign-off required
1
~70% confidence that one thing is worth scaling
This is art and science. The signal is when something is significantly outperforming everything else — an audience, a creative, a keyword — and you feel confident enough to put real budget behind it. Not certainty. Not hitting a target. Confident enough to move.
2
Minimum 2 weeks of data
Enough data to see a real pattern. Weekend vs weekday variance needs at least one full cycle.
3
You know what drove performance
You can identify whether niche, creative angle, or audience was the driver. If you can't answer that, you're not done with MVC. Unknowns pollute BAU.
4
Cam has reviewed and approved
Media buyer flags when triggers 1–3 are met. Cam reviews and approves before transition happens.
✕
Not a trigger — time alone
If 4 weeks pass with no winner, escalate to Cam. Don't auto-transition. Something is wrong with niche, angle, or offer.
Step 0
One variable. Always.
Before you read anything — check this first
A test is only valid if exactly one thing changed. If you changed the hook, the visual, and the offer at the same time — you cannot read the result. Don't try.
Before reading any test, ask: "What was the one thing we changed?"
If you can't answer that clearly, the test is worthless.
If you can't answer that clearly, the test is worthless.
Step 1
Launch on the right timeline
Know your account before you launch anything
Before launching, know whether this client's niche converts better on weekdays or weekends. Then launch accordingly to give the algorithm the best possible window.
Weekday account
Launch Monday
Weekend account
Launch Friday
Once you know which this account is, lock it in and use it every time. On a new account where you don't know yet — you're testing that too.
Step 2
Give it bare minimum 48 hours (usually 72)
Data before Day 2 is noise, not signal
Never read a test before 48 hours. Platforms need time to find their audience. Don't touch it. Don't panic. Let it run.
Step 3
Apply kill rules — but act early when signal is obvious
Juggle client-specific thresholds with the 7-day full-week rule
You don't need to wait 7 days if you launched at the right time and the signal is clear. If after 48–72 hours the ad/audience is over max cost per result — turn it off. It won't fix itself.
The 7-day window exists for when you genuinely need the full week of variance (weekday + weekend data). When it's borderline, wait and apply the client's kill rules.
Decision flow
48–72 hrs passed?
↓ Over max cost per result
Client max cost
per result hit?
per result hit?
↓ Yes
Kill it.
Move budget to winners.
Wait for next launch window.
Move budget to winners.
Wait for next launch window.
↓ Under max cost per result
Borderline or
still elevated?
still elevated?
↓
Wait for Day 7.
Apply client kill rules.
Apply client kill rules.
↓ On target
Let it run.
For client-specific thresholds (CP-Book target, kill spend, max CP-Book) — see the Kill Rules tab in the client view.
Budget rule: Always aim for 70%+ of budget on winners, 30% into new tests. Kill fast when it's obvious so budget flows to what's working.
Step 4
Compare against the winner and last 30-day data
A result only means something relative to something else
Don't read a test in isolation. Compare against the current winning ad and the last 30-day account average.
CTR
Hook strength — are people stopping the scroll?
CPC
Efficiency — what does a click cost?
CVR
Landing page — are clicks converting to leads?
CPL / CP-Book
The primary metric — cost to get a booked call.
Beats winner and 30D avg→ New winner
Beats one, not the other→ Watch
Beats neither→ Dead
Step 5
Log the result. Every time.
Win, lose, or inconclusive — it all gets logged
Every test gets logged — what was tested, what the numbers showed, what the next action is.
A kill with no log is a wasted test. Losses are data. The account log exists so the next person — or future you — knows what was already tried and why.
No log = the lesson is lost. It doesn't matter how good the result was if it's not recorded.
Core principle: CPL and CPC are market-driven. The market can double your CPM and there's nothing you can do about it. Your funnel rates are yours. The floor moves only when you can prove the funnel still works at the new cost.
When does the floor move?
Floor moves up — accept the new market reality
CPL has increased 20%+ sustained over 3+ weeks AND book rate, show rate, and close rate are all within 10% of their 90-day baseline. The market got more expensive. Your funnel is still working. Update the floor and communicate to client proactively.
Floor stays — fix the funnel first
CPL increased AND any funnel rate dropped more than 10% from baseline. This is not a market problem — it's a creative or targeting problem. Don't accept the new cost. Diagnose the funnel drop first.
Investigate before deciding
CPL increased but less than 3 weeks of data at the new level, or a campaign change in the last 2 weeks. Don't make a floor decision on noise. Wait for 3 full weeks of clean data.
Monthly baseline audit
| Metric | Pull period | Compare to | Threshold |
|---|---|---|---|
| Book rate (leads → SCH) | Last 30D | 90D avg | >10% drop = investigate |
| Show rate (SCH → Show) | Last 30D | 90D avg | >10% drop = confirmation issue |
| Offer rate (Show → Offer) | Last 30D | 90D avg | >10% drop = qualify or pitch issue |
| Close rate (Offer → Close) | Last 30D | 90D avg | >10% drop = sales problem, not ads |
| CPL | Last 30D | 90D avg | >20% increase = check funnel rates |
| CPC / CPM | Last 30D | 90D avg | Market signal only — don't act on this alone |
Client communication script
"The market has gotten more competitive in this space — CPMs are up across the board, not just on your account. The good news is your funnel rates are holding strong. Book rate, show rate, and close rate are all within 5% of your 90-day average. This tells us the system is working — the cost to feed it has just increased. We're recommending we update the CPL target from $X to $Y to reflect the new market reality and continue scaling what's working."
Lead Engine OS
Client Settings
How to use: All kill thresholds are calculated from each client's LTV. Your media buyer opens the client card, reads the numbers — no math required. Click any channel to expand its kill rules.
Lead Engine OS
Account Log
New entry
Log history