10 pm. Lights off, laptops shut, everyone gone. Your Google Ads account, meanwhile, is wide awake. Any search for what you sell can trigger an auction, at 11:47 pm just as easily as on a Sunday morning. Ads don't do office hours. The people who manage them do, and fair enough.
So we asked a simple question: what if the night wasn't dead time, but the most useful shift of the day? The answer became the way we run Google Ads with our partner roas.capital. Here is what one night looks like.
How ad spend can quietly leak overnight
Two things Google says plainly in its own help centre:
- Your daily budget is an average. On any single day Google may spend up to twice your average daily budget, while the monthly cap is 30.4 times that amount (Google Ads Help: manage your spend).
- Broad match is the default. Your ads can show for searches that are related to your keyword without containing its exact wording (Google Ads Help: keyword match types).
Neither is a bug. Both are by design. But they add up to one rule: whatever happened in your account yesterday deserves a reader today, not in next month's report, when the money is long gone.
That is where the classic agency set-up runs out of road. A retainer, a monthly call, a PDF. In between, the account runs on its own.
Log of a night shift
An illustrative example. The kitchen maker, the times and the search terms are made up for this story. No client data is involved.
- 22:00
Clocking off.
A small kitchen maker runs a campaign for made-to-measure kitchens. The workshop closes. The ads keep going.
- 23:47
A click that leads nowhere.
Someone searches “diy kitchen cabinet plans”. Broad match lets the ad in, the click gets paid for. Nobody is buying a kitchen here. They want a free blueprint.
- 01:00
The agents read the day.
AI agents go through yesterday’s data: the search terms report, costs and conversions from consent-based tracking.
- 02:30
Checking with the crowd.
Patterns are compared against anonymised, aggregated learnings from every client account. What looks like noise in one account shows up sooner across many.
- 04:00
The shortlist is ready.
Add “diy” and “plans” as negative keywords. Shift a little budget from the weaker campaign to the stronger one. Try two new ad copy variants. Every proposal comes with its reasoning.
- 07:30
One human, one coffee, one decision.
Someone from the DigiHub human layer reviews the list: approve, tweak or reject. Nothing goes live without that sign-off.
- 08:00
Test small, measure honestly.
Approved changes start as small tests against a control group. The next night they get evaluated. Whatever doesn’t hold up is rolled back.
No fireworks, no big monthly overhaul. Just one small, traceable step every night. The full method is on our Google Ads with AI page.
Monthly rhythm or night shift?
| Classic monthly rhythm | Night shift with a human layer | |
|---|---|---|
| When is data reviewed? | At the meeting, often once a month | Every night, using the previous day’s data |
| Who decides? | Whoever has time | A person, based on reasoned proposals |
| How big are the changes? | Big rebuilds in one go | Small tests with a control group |
| Where do learnings come from? | Your account alone | Plus anonymised learnings from all client accounts |
| What do you see? | A report | Every change, in your own account |
Not an autopilot. Not a black box.
“AI optimises your account” sounds like a button you press and forget. That is not how we work. Agents prepare, people decide. Strategy, goals, budget limits and exclusions are agreed with you, and the engine only moves inside those lines.
Your data stays where it belongs, in your own Google Ads account, and every change shows up in its change history. Measurement runs on first-party data, Google Consent Mode and Enhanced Conversions. No bought lists.
The interesting part is the scoring: agent proposals and human decisions are judged against the same target metric. Not to replace anyone, but to see honestly what works. ROAS is a yardstick here, not a promise.
Crowd learning instead of going it alone
Most ad accounts learn in isolation, each one making the same mistakes at a different time. With us, results flow back through our Google Ads manager account (MCC), anonymised and aggregated across all client accounts. You benefit from many accounts’ experience, and nobody gets to see your numbers.
Curious what else AI agents can take off a small business’s plate? Read our guide to AI agents for SMEs or browse the AI hub.
What does Google Ads cost with us?
- Your ad spend goes straight to Google. How much makes sense depends on your market, your offer and your goals. Quoting a one-size-fits-all number would not be serious.
- Our fee follows a simple ladder: a free audit, then a 30-day pilot at a fixed price, then a percentage of ad spend for ongoing management. Exact prices on request.
To be straight with you: advertising always carries risk. We don’t promise numbers, ROAS or rankings. What we do commit to is that your account isn’t left alone until the next monthly report.
Who is the night shift for?
- SMEs in Liechtenstein, Switzerland and Austria that already run Google Ads or are serious about starting.
- Teams who want to understand what happens in their account rather than just receive a report.
- Businesses that can track enquiries or sales, or want to set up tracking properly.
It is not for you if you are looking for guarantees, or for an account that runs itself with no humans involved. We deliberately offer neither.
Who’s watching your account tonight?
We look at your Google Ads account and your tracking and show you where we see room to improve. Free, no strings attached.
Request a free ad account auditSources
- Google Ads Help: manage your spend in Google Ads, on average daily budgets, overdelivery and the monthly charging limit.
- Google Ads Help: about keyword matching options, on broad match as the default.