Google auto-applied a recommendation in one of our accounts without my approval. The recommendation was to raise daily budgets by 25% across several campaigns "based on performance trends."
The campaigns were over-pacing. We were already going to hit budget for the month with ten days left. The recommendation accelerated that into eight days.
We burned through $40K in three days. None of it incremental. Just faster spend on the same traffic we would have gotten anyway.
That's the last time I've had auto-apply enabled on any account.
Why Google's pacing is dangerous at scale
Google's standard delivery algorithm is optimized to maximize impressions within your stated budget. The 2x daily overdelivery rule means Google can spend up to twice your daily budget on any given day if it predicts higher-than-usual traffic.
For a small e-commerce account with a $100/day budget, this is fine. Maybe fine. The overdelivery is capped at $200 and the risk is limited.
For a $650K/month portfolio with daily budgets in the hundreds or thousands of dollars, the math changes. A 2x overdelivery event across multiple campaigns in a single day can generate $20-30K in unexpected spend.
And because budget changes in Google Ads take 24 hours to fully propagate, by the time you notice and react, the damage is done.
The pacing I built instead
My custom pacing system runs against the Google Ads API daily. It does the following:
Calculates expected spend. Using weighted day math (weekends and holidays get lower weights in B2B markets), it projects what spend should look like on any given day of the month given the monthly budget and days remaining.
Compares actual vs expected. If actual spend is within 5% of expected, no action. If we're ahead of pace by more than 5%, it flags campaigns for review and optionally reduces daily budget.
Accounts for seasonal demand. Our B2B accounts have predictable seasonality — Q4 is slower because procurement budgets close out. Q1 is slower as budgets reopen. The pacing model incorporates these patterns instead of treating every month identically.
Generates daily alerts. Each morning at 7am, I get an email showing each campaign's actual vs expected spend and a pacing score. Green if on track, yellow if ahead or behind by 5-15%, red if more than 15% off pace.
Connects to pipeline data. If a campaign is running ahead of pace but also generating above-average pipeline, the system recommends keeping current budget rather than cutting. Pacing is only a problem if it's not accompanied by proportional pipeline growth.
Why not use Optmyzr or a similar tool?
Optmyzr has budget monitoring features. They work for generic accounts.
They don't understand:
- My seasonal demand model (which I've calibrated over 12 years)
- Cross-brand budget dynamics (shifting spend from Seton to Brady when Brady is underperforming)
- Pipeline velocity as a budget signal (spending ahead of pace is fine when pipeline is strong)
Generic tools understand generic accounts. I don't manage generic accounts.
The custom build took about two weekends of development time using Claude Code and basic Python. It's been running for 18 months without a significant failure.
In that time, it's prevented at least three overdelivery events that would have collectively cost $60-80K in unplanned spend. The ROI on two weekends of development is absurd.
Auto-apply is off. The system is running. That's how it stays.
Alex Langton
Senior B2B paid media manager · ~$650K/mo industrial spend
12+ years running B2B Google Ads accounts in industrial, manufacturing, and B2B e-commerce. Builds Langton Tools because generic PPC SaaS was never designed for the multi-MCC, complex- pacing, B2B-vocabulary reality of the accounts that actually drive industrial revenue.