The 1% problem: decision speed at 7-figure spend.

Most performance agencies run a rhythm that looks something like this: buyer logs in around 8–9am, scans Ads Manager, identifies anything urgent, makes decisions, goes into meetings or creative work, logs back in before end-of-day to confirm nothing broke, closes out. Repeat.

This rhythm has a built-in assumption: signals that emerge during the day can wait until tomorrow. At small spend, that's often true. At 7-figure monthly spend, it's specifically and expensively wrong.

Run the numbers. $1M/month spent 24/7/365 is $1,389 an hour. If a meaningful signal fires at 11am — say, a creative that was working stops working, or a campaign starts being rate-limited, or your conversion-tracking breaks — and you don't notice until 8am the next day, that's 21 hours of elapsed time. That's $29,000 of spend running on impaired performance.

Not every signal is a 100% waste. Most signals are “things are running 10–30% worse than they should be” rather than “things are fully broken.” So the real cost of the 21-hour lag on a typical meaningful signal is closer to 10–30% of $29,000 — call it $3,000 to $9,000 per incident.

Multiply that by even one meaningful signal a week — a very conservative estimate — and you're looking at $150K–$450K/year of spend running at impaired efficiency because your operating rhythm has an 18-hour blind window built in.

What actually works: alert-driven intraday

The rhythm that works at 7-figure spend isn't “check more often.” Checking more often is the lowest-leverage version of the fix; it just means the buyer is staring at dashboards all day looking for things that aren't there 95% of the time. That's expensive buyer time spent on false vigilance.

The rhythm that works is alert-driven: the buyer is notified when specific thresholds are crossed, and otherwise is doing deep work. Specific thresholds we've found worth alerting on at 7-figure spend:

1. Daily CPL up >25% vs 7-day averageon any campaign with >$500/day budget
2. Ad frequency >3.5on any ad with >$1,000 spend
3. Spend pacing <70% or >115% of daily budget mid-day
4. New creative crossing $2K spend with sub-threshold CTR (threshold varies by account)
5. Conversion-tracking events dropped >40% day-over-day on any campaign — usually means the pixel or CAPI broke
6. Any campaign hitting 90% of its daily budget before 3pm local — worth intraday top-up decision

The thresholds need to be tight enough that they don't become noise. At $1M/month spend we'd expect somewhere between 3 and 8 alerts on a normal day, and 15–25 on a weird day. If you're getting 50 a day, the thresholds are too loose; the buyer starts ignoring them, and you're back to the daily-rhythm problem with extra steps.

Not every 1% is the same 1%. Some gains compound through the system: better creative lowers CPM across the account, not just on one ad set; better audience signal improves Meta's algorithm globally, not just for the campaign that tuned it. Other gains are one-shot: a budget shift that rescues a day doesn't change anything about next week.

The operating implication: at high spend, prioritize the compounding 1% over the one-shot 1%, even when they're the same headline number.

Most daily-rhythm work is in the one-shot category. Most weekly and monthly work should be in the compounding category. Agencies that spend all their time on one-shot work stay treading water; agencies that protect time for compounding work build account trajectories that get better quarter-over-quarter even when nothing dramatic changes.

The specific compounding bets that paid back hardest

1. Building a real CAPI + CRM-back pipeline.On a $1M/month account, the performance gain from feeding server-side conversion data with real values is usually 8–15% on CPL within 30–45 days. That's $80K–$150K/month of improved efficiency — roughly an 8:1 return on the engineering time in year one and infinite after that.
2. Investing in a weekly creative concept review. A one-hour weekly meeting where the buyer, the strategist, and the creative lead review every concept from the prior week against data — what's working, what's dying, what patterns are emerging — usually produces a 10–15% lift in creative hit rate within six weeks. At 7-figure spend, that lift is worth more than the salaries of everyone in the meeting.
3. Running landing page A/B tests as a program, not an event.Most agencies run LP tests ad-hoc. “Let's test a new page.” At high spend, LP performance drifts under saturation the same way creative does. Running a continuous testing program — two LPs live at all times, new challenger every 3 weeks — produces a compounding conversion-rate gain that's worth several percent of the whole book.

A specific failure mode we've watched repeatedly: buyer sees a signal, isn't 100% sure what to do, waits to talk to the account manager or the client, misses the window. At small spend this caution is often correct — the cost of being wrong is much larger than the cost of waiting. At large spend the math inverts.

Rough framework we use internally:

The above isn't about skipping client communication. It's about recognizing that at high spend, an 18-hour approval loop on a pause decision for a campaign that's clearly broken has a concrete dollar cost, and that cost should be weighed against the genuine risk of acting unilaterally.

The compromise most sophisticated operators we know run is: buyers have pre-authorized decision latitude for specific classes of decisions (pause on clear anomaly; reduce budget on emerging issue; increase budget up to 20% on a clear winner), documented in the SOW, and every such action is reported same-day to the client with the rationale. This preserves the speed advantage without creating the “the agency is doing things we don't know about” problem.

A note on the labor side of this. At 7-figure monthly spend per client, the buyer managing that account is, in effect, managing $12M/year of revenue for the client. The small decisions they make daily — whether to push a budget up, whether to kill an ad, whether to escalate to strategy — directly create or destroy client revenue.

Most agency compensation structures don't reflect this. Buyer base pay is set against an industry median, not against the specific book the buyer manages. A buyer running a $5M/month book is paid roughly what a buyer running a $500K/month book is paid, with maybe a 20–30% differential for the high-spend role. That's a structural mis-pricing.

The implication isn't that buyers should all be paid more — though in many cases they should. The implication is that the buyer role at high spend needs to be supported by systems that don't rely on heroic individual judgment under fire. At $500K/month a great buyer can carry the account on their own rhythm and intuition. At $10M/month they can't — not because they're not great, but because the surface area of decisions has outgrown individual bandwidth.

This is, candidly, one of the specific reasons we built Zeke AI. Every buyer we've watched try to manage 7-figure monthly accounts at the individual level either burned out within 18 months or stopped making the high-quality decisions they made at smaller scale. The problem wasn't the buyer. The problem was that the rhythm that got them here doesn't scale there, and most agency toolstacks don't close the gap.

If you're running or managing high-spend accounts, and the 1% problem is starting to resonate:

1. Audit the lag in your daily rhythm.If signals can sit for 18+ hours before a human sees them, price that lag at roughly $1,400/hour × hours of lag × probability the signal was meaningful. You'll get a number that justifies investment in better alerting.
2. Write down your pre-approved decision latitudes. Most agencies operate on implicit norms. Writing them down and getting client sign-off turns the 18-hour approval loop into a same-day notification loop for the decisions that actually happen often.
3. Separate compounding work from firefighting work on the calendar.Protect the weekly 2-hour window for creative review and LP program management; it's the only block that produces compounding gains and it's always the first thing to get interrupted.
4. Invest in the CRM-back pipeline before anything else.Better attribution signal is the highest-ROI infrastructure investment on every 7-figure account we've audited. The platform reports are lying; your algorithm is optimizing on lies; better signal fixes that.
5. Benchmark the buyer's decision volume. How many real decisions — not reports, not meetings, not status updates — are they making a week? If the answer is in the single digits on a book worth eight figures a year, that's a structural problem worth fixing before the quarter ends.

The short version: once you're running 7-figure monthly accounts, small gaps compound into real money both ways. The rhythm that got you here doesn't scale. Build the infrastructure — alerts, CRM-back attribution, pre-approved decision latitudes, protected weekly compounding time — before you take on another account at this level.