A method that uses machine learning to set bids for each auction based on the likelihood of achieving a specified goal.
Marketing professionals and advertisers who are reading about digital advertising strategies and bidding methods.
01What it is and how it works
Automated bidding, often called Smart Bidding in Google Ads, replaces manual bid management with algorithms that adjust bids in real time for each auction. The system analyzes historical conversion data, user signals (device, location, time of day, browser, etc.), and contextual factors like search query and ad rank. It then predicts the likelihood that a given click will lead to a conversion and sets a bid that maximizes the chance of achieving your target metric — for example, a target cost‑per‑acquisition (CPA) or target return on ad spend (ROAS). The bid is set at auction time, so it can differ for every search. The machine learning model continuously learns from new data, so performance can improve over time as the system sees more conversions and adjusts its predictions.
Instead of manually setting bids, you tell the system your goal (like get more sales) and it automatically adjusts bids per auction to try to reach that goal.
02What to do about it
To use automated bidding effectively, start by ensuring you have robust conversion tracking in place. Without accurate conversion data, the algorithm cannot learn. Next, choose a bid strategy that matches your primary goal: Target CPA for cost control, Target ROAS for revenue efficiency, Maximize Conversions for volume, or Enhanced CPC for a hybrid approach. Set a realistic daily budget that gives the system enough room to bid competitively. Once the strategy is active, monitor performance daily for the first week, then weekly. Compare results against a manual bidding baseline using experiments. Adjust targets gradually — avoid large swings that confuse the model. Also, segment performance by campaign, device, and audience to see where automated bidding is working best.
03How it is measured or noticed
You measure automated bidding by looking at the metrics it was designed to optimize. For Target CPA, track actual CPA versus target; for Target ROAS, track actual ROAS versus target. Also monitor impression share, average bid, conversion rate, and cost per conversion. Use bid strategy reports in the ad platform to see how often the system changed bids and why. Run experiments that split traffic between automated and manual bidding to isolate the impact. Notice changes in performance over time: automated bidding often takes one to two weeks to stabilize after launch. A sudden drop in conversions may indicate the system needs more data or that your target is too aggressive.
04Common mistakes
- Launching automated bidding without enough conversion data — the system needs at least 30 conversions per campaign per month to learn effectively.
- Setting unrealistic targets, such as a CPA far below historical average, which forces the system to bid too low and lose auctions.
- Ignoring conversion tracking issues — if tracking breaks, the algorithm optimizes for nothing.
- Changing targets too frequently — the model needs time to adjust; daily changes prevent learning.
- Not using campaign-level budgets that align with the bid strategy — a tight budget can override the bid algorithm.
- Assuming automated bidding works the same for all campaign types — it performs best on search and shopping, less reliably on display or video.
05Limits
Automated bidding struggles when conversion volume is low (fewer than 15–30 conversions per month) or when campaigns have no historical data. It also performs poorly in highly seasonal or unpredictable markets where past data does not reflect future behavior. It cannot handle brand protection rules that require manual bid floors or ceilings. Automated bidding is often confused with simple rules‑based automation (e.g., “raise bid if conversion rate > 5%”), but true automated bidding uses machine learning, not fixed rules. It also does not replace human judgment for strategic decisions like budget allocation or creative testing.
06A worked example
An online retailer selling shoes runs a Google Ads campaign with a Target CPA of $15. After two weeks of manual bidding, the average CPA is $18. They switch to automated bidding with the same target. In the first week, the system bids conservatively and CPA rises to $22. By week three, the algorithm has learned which queries convert best and starts bidding higher on those, lower on others. By week four, the CPA drops to $14.50, below the target. The retailer also sees a 20% increase in conversion volume because the system captured more auctions at the right price.
Frequently asked questions
How is automated bidding different from manual bidding?
Automated bidding uses machine learning to adjust bids for each auction in real time, while manual bidding requires you to set bids yourself. Automated bidding aims to optimize for a specific goal like conversions or impression share, whereas manual bidding relies on your own judgment and constant adjustments.
Should I use automated bidding for a new campaign with no historical data?
It depends on your conversion volume. Automated bidding typically needs at least 15–30 conversions per month and some historical data to learn from. Without that, it may struggle and you might be better off starting with manual bidding until you have enough data.
How does automated bidding actually set bids in real time?
Automated bidding algorithms analyze signals like device, location, time of day, and user behavior at the moment of each auction. They predict the likelihood of a conversion and adjust the bid accordingly to maximize the chance of achieving your goal within your budget.
Does automated bidding still work if I have very few conversions?
No, automated bidding struggles when conversion volume is low, typically fewer than 15–30 conversions per month. Without sufficient data, the algorithm cannot reliably predict which auctions are likely to convert, leading to poor performance.
What happens if I use automated bidding without proper conversion tracking?
Automated bidding relies on conversion tracking to know what a conversion is and to optimize toward it. Without accurate tracking, the algorithm will optimize for the wrong signal or no signal at all, resulting in wasted spend and poor results.
How long does it take for automated bidding to optimize after starting?
Automated bidding usually enters a learning phase that can last 1–2 weeks, during which performance may be unstable. After that, it should stabilize and start optimizing toward your goal, but it continues to learn and adjust over time.
Asked out loud
spoken, not typedThe same term in the words somebody uses speaking to an assistant rather than typing into a box — written from the situation, which is why each one carries the situation it came from.
Yes, you can use automated bidding, often called Smart Bidding in Google Ads. It uses machine learning to adjust bids for each auction based on your conversion goals. Just make sure your conversion tracking is solid first.
The best approach is to use automated bidding, specifically target CPA or target ROAS, depending on your goal. It saves time and can improve performance, but requires sufficient conversion data to work well.
It depends on your conversion volume. If you have at least 15–30 conversions per month and some historical data, automated bidding is usually better. Otherwise, stick with manual bidding until you have enough data.