Paid Advertising
Google Ads Smart Bidding, explained without the hype
Smart Bidding is not magic. It is an auction-time algorithm that only works when it has clean data and enough remodel leads to learn from, ideally 15 to 30 a month per strategy. Here is exactly how each option works and when to use it.
Smart Bidding sets a bid for every auction using machine learning, factoring in device, location, time of day, and query intent in real time, so a homeowner searching "home addition contractor" from a desktop at night is bid on differently than a tire-kicker. Google runs this billions of times a day. But it is not a switch that prints booked jobs. Each strategy needs volume to learn, ideally 15 conversions in the trailing 30 days and closer to 30 a month for stable performance. Feed it clean conversion data and a realistic target and it earns its keep: advertisers who moved from Target CPA to Target ROAS saw a median 14% lift in conversion value at similar efficiency, per Google. Set it up wrong and it quietly burns budget.
What Smart Bidding actually does
Smart Bidding is a family of automated strategies that set a unique bid for each individual auction, not a daily average. The algorithm reads auction-time signals like device, physical location, time of day, browser, operating system, language, and the exact query, then predicts how likely that click is to convert and bids accordingly. Google performs this evaluation billions of times a day, far more granularly than any manual schedule of bid adjustments could. That is the real edge: it reacts to context a human never sees.
The catch is that prediction needs examples. The algorithm builds a model of who converts from your historical conversion data, so the quality of that data sets the ceiling on performance. If your tracking misses leads, double-counts them, or fires on the wrong action, the model learns the wrong lesson and bids against you. Smart Bidding does not fix a broken account. It amplifies whatever signal you feed it, accurate or not, which is why WellBuilt fixes conversion tracking before touching a bid strategy.
The four strategies, and what each optimizes for
There are four Smart Bidding strategies, and they split along two questions: do you have a target, and do you optimize for count or value. Maximize Conversions chases the most conversions your budget allows with no cost constraint. Target CPA adds a constraint: get the most conversions while holding an average cost per conversion you set. Maximize Conversion Value chases the most revenue, useful when leads or sales are worth different amounts. Target ROAS adds the constraint: maximize value while holding a return-on-ad-spend target you set.
Choose by what a conversion is worth to you. If every lead is worth roughly the same, like a handyman booking small repair calls, the conversion-count strategies fit. If conversions vary in value, like a $5k deck repair versus a $300k custom build that close at different rates, the value strategies win because they can sacrifice cheap, low-value leads to chase the profitable ones. Google found advertisers who switched from Target CPA to Target ROAS saw a median 14% increase in conversion value at a similar return, simply by letting the algorithm optimize the right metric.
Which strategy fits which goal:
- Maximize Conversions: most conversions, no cost ceiling, good for spending a fixed budget fully
- Target CPA: most conversions while holding an average cost per acquisition
- Maximize Conversion Value: most revenue, no efficiency ceiling, for value-varied conversions
- Target ROAS: most revenue while holding a return-on-ad-spend floor
- No-target strategies are training wheels; switch to a target once you have stable volume
Volume thresholds and the learning phase
Smart Bidding has a minimum data appetite. Google requires at least 15 conversions in the past 30 days for Search campaigns to optimize a Target ROAS or value strategy, and Display campaigns need 50 conversions with valid values in 30 days. Those are floors, not goals. Google's own guidance is to drive at least 30 conversions per month per ad group if you are sensitive to hitting your target consistently. Below the floor, the algorithm is guessing, and a campaign at one or two conversions a week may never finish learning.
Every new strategy or major change triggers a learning phase, usually about seven days but up to a month while the system tests bid levels across segments. During learning, expect your CPA to spike above target and your ROAS to dip; that is the algorithm exploring, not failing. The mistake that destroys performance is touching the campaign mid-learn. Changing the target, budget, conversion goals, or audience resets the clock and throws away progress. If volume is too low, pool campaigns into a portfolio bid strategy so the algorithm learns from their combined conversions instead of starving on each one.
Smart Bidding does not fix a broken account. It amplifies whatever signal you feed it, accurate or not.
The mistakes that quietly burn budget
The most common failure is setting an aspirational target. If your real CPA is $100 and you ask Google to hit $40, the algorithm restricts bids so hard that volume collapses and the campaign stalls in learning. Set your initial Target CPA at or slightly above your actual average from the Maximize Conversions phase, let it stabilize, then tighten 10 to 15 percent every couple of weeks. The same logic runs in reverse for Target ROAS: start near your proven return and raise the target gradually, not all at once.
The second failure is fiddling. Daily budget tweaks, target changes, and new audience exclusions each reset the learning phase, so an account that gets touched every few days never gets out of it. Give the strategy two to four weeks of clean runtime before you judge it. The third is bad inputs: counting low-intent actions like newsletter signups as primary conversions teaches the algorithm to chase the wrong people. WellBuilt audits conversion actions first so the machine optimizes toward revenue, not noise.
Why clean conversion data is the whole game
Smart Bidding is only as good as the conversions it can see. Cookie restrictions, ad blockers, and cross-device journeys hide a chunk of real conversions, and what the algorithm cannot measure, it cannot optimize toward. Enhanced conversions close part of that gap by sending hashed first-party data like email addresses back to Google to match conversions that tracking otherwise loses. Google reports a median 5% lift in measured conversions for Search after turning them on, and a median 17% for YouTube, with no change to spend.
Attribution matters just as much as measurement. Data-driven attribution spreads conversion credit across the touchpoints that actually contributed instead of dumping it all on the last click, which gives Smart Bidding a truer picture of which keywords drive results. In one Google case study, Select Home Warranty saw a 36% increase in leads and a 20% drop in cost per conversion after switching to data-driven attribution. Accurate measurement plus honest attribution is what turns Smart Bidding from a gamble into a system.
When manual bidding still beats the algorithm
Smart Bidding is not always the answer. New campaigns with no conversion history have nothing to learn from, so starting them on Manual CPC or Maximize Clicks to gather data first is often smarter than handing a blind algorithm your budget. Very low-volume accounts, niche B2B with a handful of conversions a month, can sit below the learning threshold indefinitely, where manual control or a broader portfolio is the more honest fix than a strategy that never stabilizes.
There is also a control trade-off. Smart Bidding hides the per-keyword bid, so when it misreads your goals you lose the granular levers to correct it fast. For most accounts with steady conversion volume the automation wins decisively against manual bidding's static rules. The benchmark to beat is real: the all-industry average Google Ads cost per lead sat at $66.69 in 2024 on a 6.96% average conversion rate, and a well-fed Smart Bidding strategy is usually how accounts get below it. Pick the strategy that matches your data, not the one with the best name.
Target CPA vs. Target ROAS
- Best when every conversion is worth roughly the same
- Holds an average cost per acquisition you set
- Set the initial target at or above your real average CPA
- Simpler to run for lead-gen and service businesses
- Needs at least 15 conversions in 30 days to optimize
- Best when conversions carry different values, like ecommerce
- Holds a return-on-ad-spend floor you set
- Median 14% more conversion value vs. Target CPA at similar return
- Requires accurate, differentiated conversion values to work
- Needs at least 15 valued conversions in 30 days to optimize
Key takeaways
- Pick conversion-count strategies (Maximize Conversions, Target CPA) when leads are worth the same, and value strategies (Maximize Conversion Value, Target ROAS) when they are not.
- Give each strategy at least 15 conversions in 30 days to optimize, and aim for 30 a month per ad group for stable performance.
- Set your first Target CPA at or above your real average, then tighten 10 to 15 percent every couple of weeks; never start with an aspirational number.
- Stop touching campaigns mid-learning; target, budget, goal, or audience changes reset the two-to-four-week learning phase.
- Turn on enhanced conversions and data-driven attribution before judging any strategy, since Smart Bidding only optimizes toward conversions it can actually see and credit.
SourcesGoogle Ads Help, About Smart Bidding and auction-time bidding, 2024 · Google Ads Help, Minimum conversion thresholds for Target ROAS and value-based bidding, 2024 · Google Ads Help, Set up Smart Bidding for a Display campaign, 2024 · Google for Business, Increase your ROI with value-based bidding (median 14% conversion-value lift), 2021 · Google Ads Help, Enhanced conversions median lift (5% Search, 17% YouTube), 2024 · Google Ads blog, Data-driven attribution results, Select Home Warranty case study, 2021 · Google Ads Help, About portfolio bid strategies, 2024 · WordStream / LocaliQ Search Advertising Benchmarks, 2024
Questions, answered straight.
How many conversions do I need before using Smart Bidding?
Search campaigns need at least 15 conversions in the trailing 30 days to optimize a target strategy, and Display campaigns need 50 with valid values. Those are minimums; Google recommends roughly 30 conversions a month per ad group for stable results. If you are below the floor, gather data on Maximize Conversions first or pool low-volume campaigns into a portfolio bid strategy.
Should I use Target CPA or Target ROAS?
Use Target CPA when every conversion is worth about the same, like a service business booking leads. Use Target ROAS when conversions carry different values, like ecommerce orders, because it optimizes for revenue rather than count. Google found advertisers who switched from Target CPA to Target ROAS gained a median 14% in conversion value at similar efficiency, so if you can assign accurate values, value-based bidding usually wins.
Why did my CPA spike right after I switched to Smart Bidding?
That is the learning phase, which lasts about seven days and up to a month while the algorithm tests bids across segments. CPA above target and ROAS below it are normal during this window. Do not change the target, budget, or audience while it learns, since each edit resets the clock. Wait two to four weeks of clean runtime before you judge performance.
Does Smart Bidding work if my conversion tracking is messy?
No, and that is the most expensive mistake people make. The algorithm optimizes toward whatever conversions it can see, so missed, duplicated, or low-intent conversions teach it to bid for the wrong outcomes. Fix tracking, mark only meaningful actions as primary conversions, and turn on enhanced conversions, which add a median 5% to measured Search conversions, before you trust any bid strategy.
Paid Advertising
Want this run for you, not just read about?
Make every ad dollar accountable to pipeline — lower cost per lead, sharper bidding, less waste.