Smart Bidding needs roughly 30-50 conversions per campaign, per month, before it exits the learning phase and starts optimizing well — that's Google's own published guidance, not ours. Split across too many small campaigns, none of them ever crosses that line. The fix is usually consolidation, not more budget.
- Smart Bidding needs ~30-50 conversions per campaign, per month, to stabilize — Google's own guidance, not a Rezultra number.
- Splitting spend across many small, overlapping campaigns is the most common reason bidding never leaves the learning phase.
- The audit is four checks — conversions per campaign, keyword overlap, match types, shared budgets — not a full rebuild.
We see the same account shape constantly: eleven campaigns, each one built for a slightly different product line or city, each one getting a sliver of the monthly budget. Nobody set out to build it that way — it grew one "let's just add a campaign for this" at a time. And it's usually the single biggest reason the account underperforms, not the creative, not the landing page, not the offer.
The mechanic nobody explains
Google's Smart Bidding needs data to work. Its own published guidance is that a campaign needs roughly 30-50 conversions a month before bidding stabilizes and starts optimizing well — below that, the algorithm is still exploring, not optimizing. That's a platform threshold, not a Rezultra claim, and it's the same order of magnitude Meta uses for its own ad-set learning phase.
Split that same total spend across eleven campaigns and almost none of them individually reaches that number. Bidding never leaves the learning phase. It stays cautious, spends inefficiently, and the account looks "inconsistent" month to month — which usually just means it's still guessing.
The four-check audit
This isn't a full rebuild. It's four things, checked in order, before touching a single bid.
Conversions per campaign, last 30 days. If a campaign is under ~30, it's not underperforming — it's under-fed. Judging it on ROAS at this stage is judging an algorithm that hasn't finished learning yet.
Keyword overlap between campaigns. Two campaigns bidding on the same or near-identical terms aren't reaching two audiences — they're bidding against each other, which quietly inflates your own CPCs. This is more common than it sounds once city- or product-specific campaigns pile up.
Match types in use. Broad match without a maintained negative-keyword list is the most common silent leak we find. It's not that broad match is wrong — it's that nobody's watching what it's actually matching to.
Shared vs. individual budgets. A shared budget across multiple campaigns can quietly starve your best-performing one to prop up a mediocre one, especially once campaign count creeps up. Check which campaigns are budget-capped versus which are Smart-Bidding-capped — they need different fixes.
What actually fixes it
Almost always: fewer, larger campaigns, grouped by theme rather than by every internal distinction that matters to you but not to the algorithm. Same total budget, concentrated instead of scattered, clears the learning-phase threshold faster and gives Smart Bidding something real to optimize against.
This is also exactly the kind of foundational fix we run first in our approach — audit and baseline before anything scales, because optimizing on top of a structure that's still starving for data just compounds the problem at a higher budget.
Related service: Performance Marketing