Tweaks fix accounts that work but cost a bit too much. Rebuilds fix accounts that don't work and won't, no matter how many bid adjustments you run. Three signs tell you which problem you have. If all three are true, no tweak will save it.
Sign 1 — conversions tracked, but you can't trace any back to a campaign
The account shows 80 conversions a month in Google Ads. Your CRM shows 25 new leads. The two don't reconcile, and nobody can tell you why.
The conversion tracking is firing on something that isn't a lead. Form views instead of submissions. Phone-call clicks instead of call durations. PDF downloads counted as conversions because someone forgot to filter them out.
A tweak can't fix this. The whole conversion stack has to be rebuilt from the ground — what counts as a conversion, where it fires, how it deduplicates, how it maps to revenue. That's a week of work, not an afternoon.
Sign 2 — PMax is your top spender and you've never opened its asset groups
Performance Max is 60% of your spend and you don't know what creative it's running, what audiences it's targeting, or what's inside the asset groups. The agency assured you it's "performing well", and you have no way to check.
This is the most common failure pattern we see in 2026. PMax burns budget intelligently for the algorithm and opaquely for the client. The fix isn't a setting change — it's a structural rebuild: split asset groups by service or product, set conversion values that reflect business reality, exclude brand search so PMax stops claiming attribution it didn't earn. Two weeks of work to undo nine months of drift. The shape of the diagnostic ladders up to the PMax vs Search post.
Sign 3 — CPL trending wrong for three months, nobody's stopped to ask why
Cost per lead has crept up 40% since February. The monthly report mentions it briefly and moves on to "what's planned for next month". No diagnosis. No working hypothesis. The trend is treated like weather.
Three months of unexplained CPL drift means the account has structural problems the agency isn't equipped to find. Either the audience is exhausting, the landing page has degraded, the conversion event has decayed, or the competitive landscape has shifted — all rebuild-level interventions. Tweaking bid strategies on top of a structural issue makes the chart look smoother for a fortnight, then the trend resumes.
What a rebuild means — and what it doesn't
A rebuild is a structural reset on an existing account, not deleting the account and starting over. The conversion data, audience data, search-term history, GBP links — all valuable, all kept. What changes is the campaign structure, conversion stack, asset organisation and reporting cadence.
A rebuild isn't switching to a new agency overnight. It's a deliberate 14-day window where the existing structure runs alongside the new one so we don't go dark on spend.
The 14-day window we use
- Days 1–3 : audit the existing structure, document what's broken, design the new architecture.
- Days 4–7 : build the new structure in parallel campaigns. Conversion tracking proven before launch.
- Days 8–10 : ramp new structure to 50% spend share. Old structure stays at 50%.
- Days 11–14 : if the new is matching or beating the old, ramp to 100%. If not, diagnose what's missing before fully cutting over.
The whole thing runs through the same monthly cadence we use for everything else — small, deliberate moves the client can see. That's the rhythm in the monthly report and the principle behind the wider paid media offer.
When to keep tweaking instead
If conversions reconcile to your CRM, you can see inside every campaign, and your CPL trend is flat or down — tweak. The smaller levers (negatives, bids, landing page A/B, audience refinement) are where the gains live. If the foundation is sound, a rebuild burns goodwill the account doesn't need to spend. We won't pitch a rebuild when a tweak is the right call — the same logic that runs through how we charge for ad spend.
The takeaway
Three signs, one diagnostic, one decision. If all three are true, a rebuild isn't optional — it's the only intervention that closes the gap. Novus Glass went through this in early 2025; the case study documents what came out the other side: cost per conversion down 52% in 90 days.