Why I built actcenter: the day-to-day cost of managing 25+ Google Ads accounts

Smart Bidding changed the work. Reporting tools didn't. Here's what an operator's week actually looks like across 25 accounts, and why I stopped patching it with spreadsheets.

The Tuesday that broke the spreadsheet

It was a Tuesday in March. I had 14 browser tabs open in Google Ads, a Looker Studio dashboard loading on a second monitor, three Sheets pinned to the taskbar, and Slack on a third. The week's reporting was due by Friday. The team standup was in 12 minutes.

A client had emailed at 06:48 asking why their cost-per-lead had moved 28 percent in the last week. I knew the answer was buried somewhere in those tabs. I just couldn't find it fast enough.

That morning we missed a tracking pixel that had been broken for three days. Three days of conversion volume understated. Three days of bid signal compromised. A small account, no real damage. But the operational truth landed: the surface area of the work had outgrown the tools we were using to do it.

That's the day I started building actcenter.

What the work actually looks like at 25 accounts

The sample portfolio illustrated in this post is fictional: Northpoint Digital, 24 accounts, six specialists. I built it to mirror the real shape of a mid-market paid-media operation. The numbers are made up. The pattern is not.

Operating 25+ accounts across six or seven verticals isn't 25 times the work of operating one. It's a different kind of work. It looks like this:

Five spend tiers, from the SaaS clients pushing $10K a month to the small education account at $500. Each tier has different sensitivity to noise, different cadence of decisions, different reporting expectations.

Seven verticals, each with its own keyword universe, compliance constraints, and competitive set. The roofer's worst search-term cluster is "free inspection." The dentist's is "career." The veterinary clinic's is "DIY." You can't carry one negative list across the portfolio. You have to think seven ways before noon.

Six bid strategies running in parallel. Some accounts ready for Target CPA. Some still on Manual CPC because the conversion data is too thin. Some on Maximize Conversions during a launch ramp. Each strategy has its own diagnostic vocabulary.

Daily change-history forensics. The Google Ads UI tells you what changed in the account. It doesn't tell you why it matters this week. Twenty-four accounts × seven days = 168 account-days of change to sift through every Monday.

The operator's job, end-to-end, is to take this surface area, find the three to five things that genuinely require human judgment this week, and ignore everything else without losing sleep over it.

That filter — the what matters this week filter — is the single thing every existing tool gets wrong.

Why existing tools don't fit

Looker Studio is excellent at visualizing what happened. It is not a system for deciding what to do.

Optmyzr is excellent at proposing optimizations. It is not a system for orchestrating an agency's week.

Reportei, Whatagraph, AdEspresso, Swydo — every reporting tool I tried was built around a single question: "What happened?" That's a good question. It's not the operator's first question.

The operator's first question, every morning, is: "Out of 25 accounts, which three should I touch today, and in what order?"

None of the dashboards answer that. They show you 25 things, all colored the same shade of green or red, all sorted alphabetically. Then they leave the prioritization to your judgment, your memory of last week, and the email a client sent at 06:48.

The hidden tax: 60-70 percent of time on the wrong work

I tracked my own time for six weeks before I started building actcenter. The split was:

• Reporting and data assembly: 41 percent. Pulling exports, pasting screenshots, formatting weekly decks.
• Cross-account context switching: 22 percent. Closing 14 tabs to open 14 different tabs.
• Standup and client comms: 19 percent. Repeating what the dashboard already says.
• Actual optimization decisions: 12 percent.
• Strategic / new-business thinking: 6 percent.

A team that should be spending 60 percent of its time on judgment was spending 12. The Smart Bidding era made that ratio worse, not better. Manual bidding meant the operator was the algorithm. Smart Bidding means the operator is the interpreter of the algorithm. The interpreter's job needs more thinking time, not less.

That's the day-to-day cost. Not lost spend on bad keywords. Lost human attention on the wrong work.

What actcenter changes

I built actcenter as an Operating System for paid-media operations. Not a dashboard. Not a reporting layer. A console that does three things the existing stack does not:

1. Priority Engine. The first screen an operator sees in the morning is not 25 accounts. It is three actions, ranked, with the reason each one is on the list. The 25 accounts are still there, one click away. They are not in the way.

2. Cortex Skills. Specialized AI agents for the work that used to take hours. A weekly health check that takes 8 minutes. A search-terms sweep that takes 10. An RSA refresh that takes 9. A GBP audit that takes 8. Together, the four-prompt weekly pre-flight: 35 minutes of structured pre-work, replacing 2-3 hours of report assembly. I wrote a separate post on those four prompts — read it here.

3. tCPA Health Check framework. Five states (HOLD, RAISE, LOWER, DIAGNOSE, RISK) that classify every account every week. No more "we'll keep an eye on it." Either it is HOLD, in which case you don't touch it, or it is one of the other four, in which case the next action is named. Eighteen months in production. Three agencies. Thirty-nine accounts.

What I'm not promising

actcenter does not run your account for you. The operator still owns every decision that hits production. The skills propose; the operator approves. That's not a limitation — that's the design.

actcenter does not eliminate every spreadsheet. Some workflows still belong in Sheets. We make the spreadsheets fewer, not zero.

actcenter does not work for accounts under ~$500/month in spend. Below that threshold, the conversion data is too thin for the framework to add signal.

Who this is for

If you operate 5 or more accounts, and you're spending more than half your week on reporting, this is for you. If you have a team and the quality of the work depends on which manager picked up the account, this is for you. If you've ever closed Looker Studio and opened a Google Doc to write a recommendation by hand because the dashboard didn't help you decide, this is for you.

What comes next

The next post in this series breaks down the four AI prompts I run before every weekly client call — what each one does, what it replaces, and how to install them yourself. The four skills are downloadable as standalone files at the bottom of that post.

If you operate at this scale and want to see how the system looks against your own portfolio, reach out at leoavr@gmail.com.

The work is the work. The tools should not be the bottleneck.