Quick Answer: Smart bidding doesn't run on your account's conversion data — it runs on the individual campaign's conversion data. If that campaign is generating fewer than 30 conversions per month on its own, the algorithm is guessing. Most accounts that crater after switching from manual CPC to Max Conversions or tCPA have the same root problem: not enough campaign-level signal for the algorithm to learn from. The fix is to either consolidate campaigns before switching or pool data across campaigns using a portfolio bid strategy.
Why Smart Bidding Works Differently Than Most People Think
Manual CPC works at the account level. Your experience, your keyword knowledge, your bid logic — it all applies across everything you manage. Smart bidding doesn't work that way.
When you switch a campaign to Max Conversions or tCPA, Google's algorithm looks at the conversion history of that specific campaign to build its model. It uses those past conversions to understand what kind of users, search patterns, devices, and times of day are most likely to produce another conversion. The algorithm needs enough examples to find patterns. Without them, it's essentially making random bids while it waits for data that may never come in fast enough.
This is why the "your account has plenty of conversions" defense doesn't hold up. If your account has 150 conversions a month but they're spread across eight campaigns, each campaign is starved of signal. The algorithm running campaign four doesn't know what happened in campaigns one through three. It only knows what happened inside its own walls.
We wasn't getting a ton of conversion, around 25-50 a months, but since it's high ticket, it still made sense.
We tried pmax = awful results. We tried maximize conversion with a high target CPA (500-1000$ range) = still not a lot of results.
We don't have enough conversion, therefore, the algorithms can't make sense of what we want.
These aren't unusual accounts. This is what smart bidding looks like when the data density isn't there.
The Segmentation Trap
Here's the part that trips up experienced manual CPC managers specifically: the account structure that made manual bidding work is often the exact structure that makes smart bidding fail.
Years of managing manual CPC teaches you to segment. Separate campaigns by match type. Separate by device. Separate by geography. Separate by keyword theme. That granularity gave you control over bids — you could push spend toward whatever was working and pull it from whatever wasn't.
Smart bidding does not benefit from that structure. It needs a wide view of all available conversion data to find the patterns that predict conversions. When you pre-segment by match type, you've already told the algorithm "these are separate populations" before it has any chance to see the relationships between them. Each segment ends up being too thin to inform a real model.
If you've got a high degree of segmentation in your current campaign structure — device, RLSA, matchtype, etc. — you're not really setting up the smart bidding strategy for success. Smart bidding strategies perform best when they're able to see the complete picture of data without everything already being pre-segmented.
The switch from manual to automated bidding isn't just a settings change. It often requires rebuilding the account structure first.
What to Do Before You Flip the Switch
Step 1: Check campaign-level conversion volume
Pull last 30 days of conversions at the campaign level — not the account total. Any campaign you're planning to move to smart bidding needs at least 30 conversions in that window. Fifty is better. Under 30 and the learning period will drag on indefinitely, performance will be erratic, and you'll have no way to tell whether the strategy is working or dying.
If you can't hit that threshold per campaign, you have two options: consolidate campaigns so fewer buckets are splitting the same traffic, or use a portfolio bid strategy to pool conversion data across related campaigns under a shared tCPA target.
Step 2: Set your initial tCPA at your actual current CPA
This is the single most common setup mistake. Someone running manual CPC at $80 per lead wants to hit $50 per lead with smart bidding, so they set tCPA at $50 on day one.
The algorithm reads that target and tries to find conversions at $50. If it can't — because the market doesn't support it, or because it doesn't have enough data yet to bid efficiently — it goes conservative. It stops spending. Volume dries up. The advertiser thinks smart bidding doesn't work.
Set your initial tCPA at your real current CPA or slightly above it. If you're paying $80 per lead on manual, start at $80 or $90. Give the algorithm room to find conversions before you apply efficiency pressure. Once performance stabilizes after three to four weeks, lower the target by 10-15% and hold there. Keep repeating until you reach your goal CPA.
As a general rule of thumb, your initial target for launching a new bid strategy should be for that strategy to at least match your current performance.
Step 3: Know what the learning period actually looks like
The learning period runs roughly two to three weeks. During that window, you should expect uneven performance — some days strong, some days nothing. That's the algorithm sampling different bid levels and user patterns to build its model. It's not a sign that things are broken.
What actually looks broken: zero conversions for seven or more days straight, CPAs running at three times your target with no trend toward efficiency, or impression share collapsing to near zero. Those are signs the signal isn't there, and no amount of patience will fix a data problem.
Don't make changes during the learning period. Every significant bid adjustment resets the learning phase and extends the timeline.
Step 4: Use portfolio bid strategies for low-volume accounts
If your account genuinely can't generate 30 conversions per campaign per month — which is common in high-ticket B2B and niche verticals — individual campaign bid strategies will always struggle. The fix is a portfolio bid strategy.
Portfolio bid strategies let you group multiple campaigns under one shared tCPA or tROAS target. The algorithm now sees all conversions across those campaigns as a single pool to learn from. A portfolio covering three campaigns that each produce 15 conversions a month is effectively a 45-conversion campaign from the algorithm's perspective.
Set up a portfolio bid strategy in Google Ads under Tools, then Bid Strategies. Apply it to your related campaigns. Give it the same setup care as a direct campaign strategy: start at your current blended CPA across the group, not your goal CPA.
The Bigger Picture
There's a real worldview conflict underneath all of this. Manual CPC rewards craft. The more you know about your account, your keywords, your customers, the better you can bid. Smart bidding asks you to surrender a lot of that granular control in exchange for access to signals you can't see: device combinations, user history, real-time auction context, behavior patterns across Google's network.
The trade is worth making — but only when the data is there to back it up. Accounts with consistent conversion volume, consolidated campaign structure, and accurate conversion signals tend to outperform manual CPC over time. Accounts without those things don't.
Once you get into bigger conversion numbers like 50+ the automation kicks butt.
Manual can't beat smart bidding at scale.
If your conversion volume is genuinely thin, manual CPC or enhanced CPC is not a step backward. It's the appropriate tool for your data reality.
FAQs
How many conversions do I actually need before switching to smart bidding?
The commonly cited threshold is 30 conversions per campaign per month, and it's a reasonable floor. Google has removed official minimums from their documentation, but practitioners consistently find that campaigns below 30 monthly conversions take far longer to exit the learning period and produce erratic results. If you're consistently above 50, smart bidding will almost always outperform manual over a long enough window.
Can I switch back to manual CPC if smart bidding tanks my account?
Yes, and it's not a permanent failure. If you're past the three-week learning period and still seeing CPAs at two to three times your target with no improvement, switching back to manual while you fix the underlying data problem is the right call. Save your manual bids before you switch — download a bulk sheet so you have a restore point.
My account was working fine on manual for years. What changed?
A few things happen over time. Competition increases and auction dynamics shift. Google has gradually reduced the ceiling on manual bidding's effectiveness through broad match expansion, reduced search term transparency, and auto-applied recommendations that erode negative keyword hygiene. The platform has also gotten significantly better at using signals that manual bidding simply can't access. None of that means manual CPC is dead — it means the gap between well-run manual CPC and well-run smart bidding has narrowed in smart bidding's favor at meaningful conversion volumes.
Is high-ticket B2B a bad fit for smart bidding entirely?
Not necessarily, but the data requirements are harder to meet. A B2B account with 25 conversions a month at $3,000 average deal size has fantastic economics but thin signal. The fix for most of these accounts is offline conversion tracking tied to CRM outcomes — importing qualified leads, opportunities, or closed deals as conversion events rather than form submissions. This gives the algorithm a cleaner signal to learn from and allows smart bidding to work even at lower raw conversion volumes. If you're not sure where to start with that setup, FormTrackr handles the GCLID capture and CRM-to-Google Ads pipeline automatically so you're not piecing it together with spreadsheets and cron jobs.