Google's auto-apply does not get the keys.
Auto-Apply Guard sweeps your account once a day and disables the Google recommendation subscriptions you haven't whitelisted. When Google re-enables one — and it does — we disable it again and tell you. No campaign is changed on your behalf.
A receipt of every subscription we touched.
Each daily run writes a record of what Google proposed, what was on your whitelist, and what Guard disabled. No postmortems — the answer to 'what happened to my account overnight?' is already written.
Sample · one account · one week of sweepsThe rails are boringly specific.
Guard doesn't need ML to do its job. It needs a daily sweep, a whitelist, and a log that proves what it did. Everything else is plumbing.
Runs once a day at 06:15 UTC.
For every connected account, Guard reads the current recommendation_subscription list from Google Ads, compares it against your whitelist, and disables anything outside it. Takes seconds per account.
Cadence: daily cron · every connected accountBlock all by default. Opt in per-type.
Guard uses Google's native recommendation_subscription API — the same surface Google's own interface uses. You name the categories you trust (sitelinks, callouts, broken-URL fixes). Everything else stays off, sweep after sweep.
API surface: Google Ads API · recommendation_subscriptionEvery sweep writes a run record.
What was enabled before, what got disabled, what re-enabled since last sweep — stored per-account. No rollback needed: Guard didn't change your campaigns, only revoked consent. Re-enable anything you want in Google Ads in two clicks.
Storage: tool_execution_logs · queryable per accountGuard blocked two Google-proposed budget increases that violated our lead-quality floor. The weekly postmortem we used to run is gone — I see everything the next morning.
The questions we answer on every onboarding call.
How often does Guard actually check?
Once a day at 06:15 UTC. If Google re-enables a subscription at 06:16, we catch it the next morning. If you need tighter than daily, the real workaround is to have Google stop offering the category altogether — Guard's whitelist does that for you by leaving those subscriptions permanently disabled.
Does Guard change anything in my campaigns?
No. Guard only toggles Google's recommendation_subscription flags — a consent layer that sits above your campaigns. Your bids, budgets, keywords, and strategies are untouched. Nothing to roll back, because nothing was changed on your live account.
What if Google applies a recommendation anyway, before my next sweep?
Rare, but when it happens it lands in Change History. Pair Guard with Change History Alert: Guard stops the consent at the subscription layer, Change History Alert catches the change if Google slips past it.
Can I allow some categories and block others?
Yes. The default policy is block-all. You add categories — sitelinks, callouts, broken-URL fixes, optimisation-score nudges — to your whitelist. Guard leaves those on and disables everything else, every sweep.
What happens if Guard is offline for a day?
Not much. Guard is a daily sweep, not a firewall. If the cron misses a run, your subscription state carries over from the previous day. You lose one day of protection, not the account. The next successful sweep re-asserts the whitelist.
Run the audit.
Keep the findings.
Connect your account and let Lyra run its 18 tools for 14 days. If the projected waste recovery isn't worth at least 10× the $49, don't pay. You keep every insight either way.
Read-only connect · write access opt-in per tool · SOC 2 in progress · GDPR + CCPA compliant