If performance tanked right after you increased budget and applied support/auto suggestions, don’t keep tweaking live. Stabilize first: use Change History to revert risky edits, restore relevance (structure/ads/negatives), verify tracking, and ramp budget gradually. Test anything new via Drafts & Experiments.
What usually goes wrong
- Budget shocks reset learning and push you into auctions you can’t win yet.
- Auto-applied/support suggestions often loosen targeting (broad match, audiences, locations) or flip bidding strategies abruptly.
- Ad/landing misalignment tanks CTR → lowers QS → raises CPC → kills volume.
- Tracking drift (new conversion actions, value rules, or duplicates) misguides smart bidding.
Immediate triage mindset
- Objective: restore baseline (the last known good state), not “try 10 new ideas.”
- Work in a copy/experiment whenever possible. Limit live changes to reversions.
Step-by-step recovery plan
1) Change History deep dive (last 30–90 days)
Export changes by type & date: budgets, strategies (Max Conv ↔ tCPA, Max Conv Value ↔ tROAS), match types, RSA rewrites, negatives removed, audiences, locations, ad schedule, PMax asset changes. Tag each as low/med/high risk.
2) Roll back high-risk edits first
- Revert sudden bidding flips (e.g., to tROAS with thin data).
- Re-tighten match types (undo unnecessary broad, bring back exact/phrase).
- Restore negatives that were removed.
- Revert location/schedule expansion that spiked waste.
- If RSAs changed and CTR fell, restore the best-performing variants.
3) Rebuild structure & relevance
- Keep 1–3 close variants per ad group (or SKAG-style if you prefer).
- Mirror the keyword in H1 and Path.
- Align landing pages with the promise/intent; keep forms above the fold.
- Re-add extensions/assets (sitelinks, callouts, structured snippets).
4) Verify conversions & signals
- Confirm the primary conversion action (not secondary events).
- Check for duplicates or action swaps, and validate tag firing.
- Ensure value settings and enhanced conversions are correct.
5) Budget ramp plan (avoid shocks)
- If you must scale, +15–25% per step, then hold 5–7 days.
- Watch Conv, CPA/ROAS, IS (rank), CTR, QS comps, and Search Terms.
6) Smart bidding sanity checks
- tCPA/tROAS setpoints must match reality. If over-tight, loosen to regain volume.
- For Maximize strategies, confirm sufficient daily budget and signal quality.
7) Search Terms & negatives (daily at first)
- Rebuild negatives for off-intent queries; maintain shared lists (careers, free, DIY, competitors if needed).
8) Audiences, locations, schedule
- Confirm location targeting is “Presence” (not presence or interest, if that hurt you).
- Trim underperforming hours/days; protect peak windows.
9) Use Drafts & Experiments (D&E) for anything new
- Test broad match, bid strategies, new RSA assets, or geos in Experiments, not the main campaign.
- Let experiments run 2–4 weeks with enough spend for significance.
10) Stabilize → then scale
- Only increase budgets again after stable KPIs (≥7–14 days) and clear lift.
Quick checklist
- Change History exported & high-risk edits listed
- Risky changes reverted (bids, match, negatives, geo/schedule)
- Structure tight; ads mirror keywords; LPs aligned
- Conversions verified; enhanced conv/value rules correct
- Budget ramp staged (+15–25% steps)
- D&E for new ideas; no live gambles
- Daily Search Terms & negatives for 1–2 weeks
- Monitor learning periods and lag before judging
FAQ
Should I revert everything?
No—only high-risk or clearly harmful edits. Keep improvements that align with your prior winning setup.
How long to recover?
Expect a learning period and lag; watch leading indicators (CTR, Search Terms quality, IS) before conversions catch up.
Is broad match bad?
Not inherently—just test it in Experiments with strong negatives and solid signals.
Need a second pair of eyes?
We run audits and full management as a Google Partner. Get in touch and we’ll map a recovery sprint for your account.