Quick Answer: Smart bidding needs conversion data to learn, and accounts under roughly 30 conversions in 30 days rarely give it enough. If you switch a starved campaign to tCPA or max conversions, impressions often collapse. The fix is a decision path. Pool similar campaigns under a portfolio bid strategy, feed the account more conversion signals through offline conversion import, or hold on max clicks until volume supports automation.
The low-volume trap
You run a lead gen account in a narrow niche. Maybe B2B, maybe a local service with high ticket prices. The account brings in 5 to 10 leads a month. You know smart bidding can outperform manual CPC because you have seen it work in bigger accounts. So you move one campaign over to tCPA or max conversions.
Then impressions fall off a cliff. The campaign sits in the learning phase for days. You check search results for your own keywords and your ads are nowhere. After a week of near-zero spend you retreat to manual CPC and wonder if Google's automation was ever built for accounts your size.
Here is the uncomfortable truth. It mostly wasn't. Smart bidding is a prediction engine. Predictions need historical conversion data, and 8 leads a month is not a dataset. That does not mean you are stuck forever. It means you need a plan that matches your volume instead of pretending you have an ecommerce account doing 500 orders a week.
First, check your data floor
The number people repeat is 30 conversions in the last 30 days. Treat it as a rough floor, not a magic switch. Two things matter more than the raw count.
The conversions have to live inside Google Ads. Organic form fills, conversions recorded only in GA4, and app events sitting in Firebase do nothing for smart bidding unless they are imported as conversion actions the bid strategy can optimize toward. Plenty of advertisers think they have volume when the ad account itself sees almost none of it.
The conversions also have to be recent and consistent. Thirty conversions spread across seven months tells the algorithm nothing about the current auction. If your last 30 days are thin, your history will not save you.
Count your primary conversion actions inside the ad account over the last 30 days. That number decides which branch of the framework you take.
The decision framework
Step 1: Pool data with a portfolio bid strategy
If you run several campaigns that each get a handful of conversions, stop making the algorithm learn each one in isolation. A portfolio bid strategy lets multiple campaigns share one bidding model, so five campaigns at 8 conversions each become one strategy learning from 40.
The catch is similarity. Group campaigns that target the same kind of buyer, drive the same conversion action, and carry a comparable CPA. A portfolio full of mismatched goals confuses the model instead of feeding it. Think one product line across several cities, not lead gen bundled with ecommerce.
One warning from advertisers who run these. Google tends to shift budget toward whichever campaigns convert most easily and let the rest starve. If every campaign in the portfolio matters equally to the business, watch the distribution weekly.
Step 2: Add conversion signals with offline conversion import
Low-volume accounts often have more conversions than Google ever sees. Phone calls that turn into booked jobs. Form leads that become sales two weeks later in the CRM. None of that reaches the bid strategy unless you send it back.
Offline conversion import closes the gap. Capture the GCLID when a lead comes in, then upload the qualified outcomes as conversions. You get two wins. The account gains first-party conversion volume it was already earning, and the algorithm learns from your best outcomes instead of raw form fills. For many niche accounts this is the difference between starving smart bidding and clearing the floor.
Step 3: Hold on max clicks and fix structure
If you cannot pool and cannot import, do not force tCPA. Run max clicks or manual CPC while you build volume, and check whether your own structure is choking impressions. Exact match only plus desktop only plus a tight geo is a common setup in small accounts, and it can shrink the auction so much that no bid strategy has room to work. Open one variable at a time and manage negatives closely.
Raising a too-low target also matters. When a campaign on tCPA refuses to spend, the target is often set below what the auction actually costs. But in a genuinely starved account, a higher target only buys you a slower failure. Volume comes first.
Step 4: Re-check monthly, switch once, then leave it alone
When your 30-day conversion count clears the floor, switch one campaign or one portfolio and give it two to three weeks without edits. Every significant change resets learning. The accounts that fail at smart bidding twice usually failed because someone kept touching the campaign every three days.
Most of this framework depends on one thing. Conversions actually making it into Google Ads. If your calls, CRM outcomes, or qualified leads never get uploaded, your account looks smaller than it is and every bid strategy underperforms. The free OCT Setup Checklist walks through the exact offline conversion tracking setup, from GCLID capture to the upload format Google accepts. Grab it at freak.marketing/oct-checklist.
What not to do
Do not toggle between bid strategies every few days looking for the one that spends. Each switch restarts learning and burns budget on nothing. Do not trust that Google will lean on industry benchmarks to carry a new tCPA campaign. Some advertisers report it working, but the results are unreliable and you have no visibility into what the model assumed. And do not conclude that better leads would fix everything. If your CPA math does not work at your ticket price, no bid strategy will rescue the account.
FAQs
Does the 30 conversions in 30 days rule come from Google?
Google has published shifting guidance over the years, and current products will let you enable tCPA with far less. The 30 in 30 figure survives because practitioners keep seeing accounts below it stall in learning. Treat it as a practical floor based on results, not a documented requirement.
Do GA4 conversions count toward smart bidding volume?
Only if they are imported into Google Ads as conversion actions and set as primary. Even then, ad-attributed conversions inside the account carry the real weight. Conversions that happen organically do not teach the bid strategy anything about your ads.
Will a portfolio bid strategy hurt my strongest campaign?
Usually the opposite happens. The strong campaign gets more budget and the weaker ones get starved. If a strategically important campaign is losing spend inside the portfolio, pull it out and let it run its own strategy, or set impression share protection where coverage matters more than efficiency.
How long should I wait before judging a smart bidding test?
Two to three weeks minimum, with zero major edits during that window. Judge it on cost per conversion across the full window, not day-by-day swings. If impressions are still near zero after the first week, check your target and your match type restrictions before blaming the algorithm.