How to fix: Smart Bidding running under 30 conversions per 30 days

TL;DR

Your campaign uses a Smart Bidding strategy (tCPA, tROAS, or Maximize Conversions) but reports fewer than 30 conversions in the last 30 days. Below that floor the algorithm cannot stabilize, so CPAs swing wildly and budget is spent on exploration instead of converting auctions. Fix: either raise the conversion signal (Enhanced Conversions, Offline Conversion Imports, broader conversion action set, more spend) or step back to Manual CPC / Maximize Conversions without a target until the campaign clears the threshold.

Why it matters

Smart Bidding strategies are machine-learning systems. They observe historical conversions, learn which auction signals correlate with conversion probability, and then bid in real time per query. The model needs a minimum data density to do this — Google publishes 30 conversions in the last 30 days as the practical floor for performance evaluation across tCPA and Maximize Conversions [1] [2]. Target ROAS adds a stricter gate: the conversions must also report revenue values, because value-based bidding cannot calibrate against zero-value events.

Below that floor the algorithm is permanently in exploration mode. Each new query gets a wide-band bid because the model has too few prior observations to confidently estimate value. The visible symptom is week-over-week CPA variance of 20-40%, scattered spend on irrelevant search terms, and a "Learning" or "Learning (limited)" status that never clears. Smart Bidding will not produce a target-CPA outcome here — it will produce an average over a wide distribution that includes a long tail of overpriced clicks.

The reason this persists in real accounts is incentive misalignment. Account managers turn on tCPA because Google's UI nudges toward it, and Google's auto-recommendations actively promote the switch. But Google's recommendation engine does not gate on conversion volume — it will surface tCPA on a campaign with 4 conversions per month, where the strategy is mathematically guaranteed to fail. The result is a sub-threshold campaign running a strategy designed for a higher-volume regime, burning learning budget indefinitely.

The recoverable spend is meaningful. On audits of mid-market accounts, sub-threshold Smart Bidding campaigns typically run 30-60% above target CPA during the first 60-90 days, and never converge. Switching to Maximize Conversions without a target (or Manual CPC) holds CPA at the campaign's actual market rate and avoids the volatility tax.

How to fix

1. Check the campaign's 30-day conversion count. Open the campaign Overview tab (Campaigns → select campaign → Overview) and read the Conversions column over the last 30 days. If the count is below 30 — or below 30 with revenue value when on tROAS — the gate is not cleared.
2. Switch the bid strategy to a volume-agnostic option. Go to Settings → Bidding → Change bid strategy. For most sub-threshold campaigns, Maximize Conversions without a target works best — it still uses Smart Bidding signals but does not enforce a target the algorithm cannot honor. For very low volume (<10 conversions/month), drop to Manual CPC and accept that you are bidding without ML lift until volume grows.
3. Raise reported conversion volume before re-escalating. Three levers, in order of effort: enable Enhanced Conversions on the conversion action (Tools → Conversions → select action → Enhanced conversions section → turn on) to recover 3-15% of conversions lost to tracking gaps [3]; add Offline Conversion Imports via Data Manager [4] if you run a CRM with GCLID capture; broaden the primary conversion set to include strong lead actions (e.g. demo request and qualified-lead-imported-from-CRM together) provided each action is genuinely revenue-correlated. See Enhanced Conversions setup and conversion tracking foundation for sibling fixes.
4. For tROAS specifically, verify revenue values are populated. Open the conversion action and confirm each conversion ships a non-zero value field. If half your conversions carry value=0, the algorithm sees them as worthless and skews bids accordingly. Either populate value at conversion time, or remove value-less actions from the primary set when running tROAS.
5. Re-evaluate after 30 days, not 7. Smart Bidding's learning phase typically lasts 7-14 days [1], but conversion-count stability needs a full 30-day window. Only escalate back to tCPA or tROAS after the campaign sustains ≥30 conversions for two consecutive 30-day windows — one window can be a seasonal spike.

Common mistakes

• Setting a tCPA target that is far below current CPA on a sub-threshold campaign. The algorithm cannot lower CPA when it cannot even predict CPA. The result is a near-zero impression share, not a lower CPA.
• Switching strategies every two weeks. Each strategy change triggers a fresh 7-14 day learning phase. Three changes in a month means the campaign spends the entire month learning and never optimizing.
• Aggregating campaigns into a portfolio just to clear the threshold. Portfolios work when the constituent campaigns share conversion mechanics. If you portfolio together a brand campaign and a cold-prospecting campaign, the algorithm averages bid behavior across two very different funnels and degrades both.
• Enabling Auto-Apply recommendations to "automatically optimize" a sub-threshold campaign. Auto-Apply will accept the tCPA recommendation Google surfaces, locking you into the failing strategy. Keep recommendations under manual review.
• Confusing 30 micro-conversions with 30 macro-conversions. Counting page-views or scroll-depth toward the 30/30d gate is a fiction. Only actions that reflect business value (lead, signup, purchase) count for bidding-quality purposes.

FAQ

Is 30 conversions per 30 days a hard rule or a soft guideline?
It is Google's documented evaluation floor for Smart Bidding [1] [2]. The algorithm will technically run below 30 conversions, but the predictions are unreliable. Treat it as the practical minimum, not as a soft guideline.

Does Target ROAS need the same 30 conversions?
Google's tROAS documentation lists 15 conversions in 30 days at the conversion-tracking level as the published minimum for Search and Shopping [2]. In practice, account audits show tROAS stabilizes around 30 conversions with reported revenue value. Treat 30/30d with value as the reliable working floor; 15/30d is the floor where tROAS will at least run, not where it performs.

Can I use Maximize Conversions with a low target CPA option instead?
Maximize Conversions with a target CPA behaves like Target CPA [1] — same volume gate applies. Maximize Conversions without a target is the sub-threshold option.

My campaign just launched — should I start on Manual CPC or Smart Bidding?
Start on Maximize Conversions without a target if conversion tracking is fully wired and the conversion action is realistic for the budget. If conversion tracking is shaky or the action is rare, Manual CPC with tight match types for the first 30 days, then re-evaluate.

What if I cannot raise conversion volume — budget is fixed?
Stay on Maximize Conversions without a target indefinitely. It is not a "downgrade" — it is the correct strategy for a sub-threshold campaign. Resist Google's recommendation to "improve performance" by switching to tCPA; that recommendation does not gate on conversion volume.

Sources

1. Google Ads Help — About Smart Bidding. Defines the four Smart Bidding strategies, the evaluation floor (≥30 conversions in 30 days for tCPA / Maximize Conversions; ≥15 conversions in 30 days at the conversion-tracking level for Target ROAS on Search and Shopping per published minimum), and the learning-phase behavior (7-14 days) during strategy transitions.
2. Google Ads Help — About Target CPA bidding. Establishes the recommended measurement window of 30 days and at least 30 conversions for tCPA evaluation.
3. Google Ads Help — About Enhanced Conversions. Documents how hashed first-party data sent via the Google tag improves conversion measurement accuracy, recovering events lost to consent and tracking gaps.
4. Google Ads Help — Offline Conversion Imports via Data Manager. Describes GCLID-based offline conversion uploads from CRMs (Salesforce, HubSpot, sFTP, Google Sheets) into Google Ads.