Manual bidding and Automated bidding at a glance
With manual bidding, the advertiser sets maximum cost-per-click (CPCAverage cost of each click.) bids and adjusts them using reports: raising bids on keywords that convert, lowering them on ones that do not, and adding adjustments for devices, locations or schedules. Every change is visible and every decision is a person's. The limitation is that a bid set at the keyword level applies to many different auctions with different odds of converting, and a person can only adjust in broad strokes and at intervals.
With automated bidding, the advertiser sets a goal, such as maximising conversions, a target cost per acquisition (CPACost of each conversion.) or a target return on ad spend (ROASRevenue generated per unit of ad spend.), and the platform sets the bid for each auction. The Google Ads help centre documents how its Smart Bidding strategies use auction-time signals, including device, location, time of day, language and audience lists, to predict the chance of a conversion. Other platforms describe similar systems. The algorithm adjusts constantly, but it can only optimise towards the conversions it is told about.
This page decides between the two control styles. The individual strategies and how to configure them belong to the Smart Bidding strategies guide, and the choice between the two most common goal types is covered in the target CPA versus target ROAS comparison.
Compare table
Manual bidding vs Automated bidding
| Row label | Manual bidding | Automated bidding |
|---|---|---|
| Who sets bids | The advertiser, per keyword, product group or audience | The platform, per auction, towards a goal the advertiser sets |
| Data required | Works with little data, relying on the manager's judgement | Needs accurate conversion tracking and steady recent volume |
| Signals used | What the manager can see in reports and set as adjustments | Many auction-time signals the manager cannot see or set individually |
| Control | Direct control over the maximum price paid | Control through goals, targets, budgets and data quality |
| Transparency | Every bid and change is visible | Bids per auction are not shown; reporting shows outcomes and some signals |
| Learning period | None: changes take effect at once | A period after launch or major change while the strategy calibrates |
| Response to change | As fast as the manager notices and acts | Fast for patterns in the data; slow or wrong for sudden shifts it cannot see |
| Best-fit accounts | New, small or low-conversion accounts; tightly capped placements | Accounts with reliable tracking and steady conversion volume |
| Failure modes | Missed opportunities, slow reactions, time-consuming upkeep | Optimising to bad conversion data, overspending to meet volume, opaque drift |
The rows on data and failure modes matter most. Automated bidding is only as good as the conversion data it learns from, and when it fails it tends to fail quietly.
When to choose Manual bidding
Manual bidding is the sensible choice for a new account or campaign that has no conversion history. An algorithm with nothing to learn from either bids cautiously and spends little, or bids widely and wastes budget while it explores. Manual bids let the advertiser gather clicks and conversions at a controlled price until there is enough history to hand over.
It also suits campaigns that will never produce much data. A niche business-to-business campaign with a handful of conversions in a typical month, or a local service with a small search market, may never give an automated strategy a stable signal. In those cases a manager watching search terms and adjusting bids by hand often outperforms an algorithm guessing from sparse data. Before settling on manual bids for that reason, the guide to bidding with few conversions sets out ways to give an automated strategy more to learn from, such as consolidating campaigns or bidding on a more frequent action.
Manual control remains useful where the price ceiling matters more than volume. Brand terms, where the advertiser wants presence at a low cost, and placements with strict caps are examples. Finally, manual bidding is the safe fallback when conversion tracking is broken or under review. Automating on top of bad data teaches the algorithm the wrong lesson; fix the tracking first, using the conversion tracking setup guide, then decide.
When to choose Automated bidding
Automated bidding is the stronger choice once an account has reliable conversion tracking and a steady flow of conversions. At that point the algorithm can use signals a manager never sees at auction time, and it adjusts every auction rather than a keyword's average. In most well-tracked accounts with sufficient volume, that granularity wins.
It suits accounts that care about value rather than just volume. When conversion values are passed to the platform, a value-based strategy can bid more for auctions likely to produce larger orders or more valuable leads, which manual bidding cannot do at the same resolution. It also suits complex accounts where the number of keywords, products and audiences makes manual management impractical, freeing time for work the algorithm cannot do: offers, creative, landing pages, feeds and negative keywords.
The conditions matter as much as the choice. Set targets close to recent actual performance, not to an aspiration, because a target far from reality either starves the campaign or pushes spend into poor auctions. Keep budgets and targets stable through the learning period. And watch the inputs closely: an automated strategy is now managed through the quality of its conversion data.
Decision table
| Situation | Choose | Why |
|---|---|---|
| New account or campaign with no conversion history | Manual | The algorithm has nothing to learn from |
| Reliable tracking and steady conversion volume | Automated | Auction-time signals outperform keyword-level bids |
| Only a few conversions in a typical month | Manual, or an automated clicks goal with caps | Too little data for conversion-based learning |
| Conversion tracking broken or under review | Manual until fixed | Automation would learn from bad data |
| Conversion values differ widely between orders or leads | Automated, value-based | Bids can reflect predicted value per auction |
| Brand terms where low cost matters more than volume | Manual or a capped strategy | Direct control over the price ceiling |
| Large account with many keywords and products | Automated | Manual management cannot keep pace |
How to switch safely
Treat a switch as a test, not a setting change. The most reliable way to judge an automated strategy is to run it against the current approach on comparable traffic. Platforms offer campaign experiments that split traffic or auctions between two bidding approaches, and a geographic split gives a cleaner read when the experiment tools do not suit the account; the bidding strategy geo split test sets out a ready-made design with a hypothesis, metrics and guardrails.
Before switching, check the conversion actions. Make sure only the actions that matter are set as primary goals, that duplicates are removed, and that values are passed where the strategy uses them. Then set the target from recent actual performance, not from the hoped-for figure, and plan to tighten it gradually.
During the test, leave it alone. Changing budgets, targets or conversion settings mid-test resets learning and muddies the read. Decide the review period in advance and judge the result on the business's own outcome, such as cost per qualified lead or contribution after advertising cost, rather than only on the platform's reported conversions.
After switching, keep watching. Automated strategies drift when the market, the site or the tracking changes. Monitor cost per conversion, conversion rate, search terms and spend against plan every week, and keep a documented route back to manual control if tracking breaks.
A home services firm tests automation on two regions
Clearflow Plumbing runs search campaigns in six regions on manual bids, averaging 140 booked jobs a month at a cost per booking of about £38. Its tracking was rebuilt last quarter to count only confirmed bookings. It moves three regions to a target cost per acquisition strategy with a target of £38, and keeps three on manual bids, matched by past volume. After six weeks, the automated regions record 12% more bookings at a cost per booking of £36, while the manual regions are flat. The firm moves the remaining regions to automation, keeps brand terms on a separate capped campaign, and schedules a monthly check of the booking data the strategy relies on.
Common mistakes
- Automating on top of bad tracking, so the algorithm learns to buy the wrong conversions.
- Setting an aspirational target far from recent results, then blaming the strategy when volume collapses.
- Changing settings during the learning period, resetting the strategy before it settles.
- Switching back and forth after a few bad days, so neither approach gets a fair read.
- Judging on platform-reported conversions alone, without checking the business's own records.
- Assuming automation means no management. The work moves to data quality, targets, feeds and creative.
- Keeping manual bids out of habit on a well-tracked, high-volume account where automation would do better.
