More insights.
The learning never stops.
Strategy, creative and numbers feed each other every week. What the loop teaches us lands in your inbox.
Most agencies are managing budgets. The job is reading signals.
The standard description of agency work on Meta: upload ads, manage budgets, optimise delivery, report ROAS. We do all of that, mechanically, every week. It's table stakes. The actual job is reading what Meta tells you every day and making the next decision from that information instead of from memory.
The algorithm has more data than any media buyer ever will. What it doesn't have is taste, context, or a creative strategy. What it has is feedback: billions of micro-signals about which ads work for which people. Reading that feedback is the work. Everything else flows from what the reading says. Here's the whole operating rhythm, with the actual numbers, so you can run it on your own account.
Rule one: upload on a calendar, not a reaction
New ads go up every Friday. Without exception, regardless of what last week's data said. Most agencies are reactive: they upload more when things go well and freeze when things go badly. The cadence is the system. Stop feeding Meta and the account stalls: winners fatigue, frequency creeps, CPMr rises, and CPA follows two to six weeks later, by which point the damage is already in motion.
Three to ten new ads per brand, every Friday, each briefed against a specific persona, awareness level, angle and format, each named by convention so that six months later you know exactly what was tested. Every variable intentional, which is what makes the data that comes back interpretable.
The 15-minute morning read
Every weekday, every account gets read. Not dashboard-checked. Read. A dashboard says ROAS is 2.4 and spend is on pace. A read says Meta shifted 30% of spend to a new ad in 48 hours, that ad targets an untested persona, and the old top spender's CPMr just rose for a second straight week, so the account's most important creative for the next month just changed. Three checks:
1. Spend allocation. Where Meta places spend is market feedback. An ad taking 20%+ of total spend is a top contributor. Under 10%, the system is still learning. Watch the shape: which ads are rising, fading, or suddenly climbing from dormancy. Spend moves before conversion metrics do.
2. CPMr. CPM × frequency: what you actually pay to reach a unique person. ROAS is a lagging indicator; by the time it drops, the damage happened weeks ago. CPMr leads. Two consecutive weeks of rising CPMr means CPA deteriorates within six weeks. The response is never more budget. It's new creative against a different persona, briefed before the decline starts.
3. Bidding model behaviour. Each model has a decision ladder, not a vibe. A Cost Cap spending 75-100% of its daily budget at 2+ ROAS: raise the bid, unlock volume. A Cost Cap spending 0-25%: the creative isn't working, and no bid will fix it. Brief new creative. In between, hold and watch. The ladder makes ten decisions an hour possible, with confidence, because none of them are improvised.
The honest numbers
Meta inflates its reported ROAS. Last-click and view-through attribution double-count, claim credit for purchases the ad didn't drive, and never subtract refunds. The number in Ads Manager is not the number to manage the business by.
The only honest ROAS is total Shopify revenue divided by total paid spend across all channels. A brand at 4.2 on Meta's dashboard is often at 2.6 blended. Those are different businesses. Set your reporting against Shopify ground truth and make scaling decisions from the blended number, because that's the one your bank account agrees with.
Alongside it, track NCCPA: total paid spend divided by new customers (orders_count = 1 in Shopify). Great ROAS with a rising NCCPA means Meta is recirculating existing buyers. The numbers look healthy while new customer acquisition quietly dies, and the plateau arrives six months later. NCCPA catches it first.
The system that does the collecting
Reading seven accounts manually took three hours a morning, so in 2026 we built an in-house intelligence system. Every day at 8am it pulls live Meta and Shopify data per brand, calculates CPMr at account and ad level, spend zones, ATC% against the brand's 30-day rolling average, blended ROAS against Shopify ground truth and the new-versus-returning split, then posts a two-part briefing to Slack: the data, and a plain-English strategic read. Report generation that took an afternoon now takes a Slack command and sixty seconds. The point isn't the automation. It's that the saved time gets spent on interpretation, the part that changes outcomes.
What each reading triggers
The read produces the move. The moves are rules, not moods:
CPMr up two consecutive weeks: brief a different persona or awareness level from the strategy map. The map guarantees unused directions exist.
A new ad emerging into the 10-20% spend zone with strong ATC%: compound it. Same hook on a new body, same angle in a new format, five hook variations. The winner becomes a seed, not a thing to ride until it dies.
Top spender stable on spend share but CPMr rising: fatigue is arriving. Prepare the replacement now. The pipeline stays six weeks ahead of current performance.
NCCPA rising while ROAS holds: a new-customer problem masked by returning-customer behaviour. Shift briefs to unaware and problem-aware angles and go find new people.
Meta-first, on purpose
We run Google where the search intent genuinely exists, and route spend to whichever channel wins the head-to-head. But the agency is built around Meta: the creative process assumes its logic, the Friday cadence is timed to its auction, the bidding ladders are specific to its products. Twelve years in the same auction, through four algorithmic shifts and two privacy overhauls, produces pattern recognition that a channel-agnostic generalist can't replicate. Saying so is more honest than pretending otherwise.
The principle
The platform is the brand's market, talking, in a vocabulary of spend allocations, CPMr trends, frequency thresholds and bidding behaviour. Upload every Friday. Read every morning. Manage to the blended number and NCCPA, not to Meta's fiction. Let every reading trigger its rule. The agency that listens beats the agency that doesn't.
If you want to see this applied to specific brands, the work is here. If you want us to run it with you, book a call.