Automated bidding is the process by which an advertising platform’s AI sets a bid for every auction automatically, adjusting in real time to hit a goal you define, whether that is conversions, revenue, visibility, or clicks. The single most important thing to understand upfront: the algorithm does not just set one bid and leave it. It recalculates at every auction, factoring in dozens of signals about the user, the context, and your account’s performance history. Platforms including Google Ads, Microsoft Advertising, and Google’s Smart Bidding subset all operate on this principle, though they name their strategies slightly differently.
Key takeaways
Automated bidding works best when conversion tracking is accurate, targets are realistic, and account structure gives the algorithm clean data to learn from.
| Point | Details |
|---|---|
| Definition | Automated bidding uses AI to set a unique bid per auction, optimising toward a goal you define (conversions, ROAS, clicks, or visibility). |
| Strategy mapping | Match your goal to the right strategy: Target ROAS for ecommerce revenue, Target CPA for lead generation, Target impression share for visibility. |
| Prerequisites | Reliable conversion tracking, correct attribution, and sufficient conversion volume (30–50 per month minimum for CPA/ROAS strategies) are required before switching. |
| Learning period | Expect 1–4 weeks of learning; avoid structural changes during this time and monitor the bid strategy status report daily. |
| Oxedent | Oxedent manages Smart Bidding strategy, tracking setup, and feed optimisation for ecommerce brands spending £2,000+ per month on paid media. |
Table of Contents
- Which automated bidding strategies are available and what does each one optimise for?
- How automated bidding actually works: signals, auctions, and learning
- When automated bidding is the right choice (and when it is not)
- Setting up automated bidding correctly: a practical checklist for UK advertisers
- Common problems after switching to automated bidding and how to fix them
- Three practical scenarios: matching goals to the right strategy
- An agency perspective on automation and where human strategy still matters
- Oxedent’s ecommerce PPC management: built around profitable automation
- Sources
Which automated bidding strategies are available and what does each one optimise for?
Understanding automated bidding strategies means mapping each one to a specific business goal. Pick the wrong strategy and the algorithm will optimise hard for the wrong outcome. Here is a clear breakdown of the main options, drawn from Google Ads’ official strategy definitions.
1. Target CPA (cost per acquisition)
The algorithm sets bids to get as many conversions as possible at or below your target cost per acquisition. This is the right choice when you know your acceptable cost per lead or sale and have enough conversion data for the model to work from. Typically, you need a sufficient number of conversions in the past period before the strategy performs reliably.
2. Target ROAS (return on ad spend)
Rather than optimising for conversion volume, Target ROAS optimises for conversion value relative to spend. If you sell products at varying price points, this strategy tells the algorithm to chase higher-value transactions more aggressively. For ecommerce brands, this is often the most commercially meaningful strategy once value tracking is in place. Practical guidance on improving ROAS for ecommerce is worth reading alongside this.
3. Maximise conversions
No target constraint. The algorithm spends your full budget to drive as many conversions as possible, regardless of cost per conversion. Useful for new campaigns building data, or when volume matters more than efficiency. Adding a Target CPA cap later constrains it toward a cost goal once data accumulates.
4. Maximise conversion value
The value-focused equivalent of Maximise conversions. The algorithm prioritises higher-value transactions without a ROAS floor. Again, adding a Target ROAS constraint later shifts it from pure volume to value efficiency.
5. Maximise clicks
No conversion data required. The algorithm bids to get as many clicks as possible within your budget. Suitable for awareness campaigns, new accounts with no conversion history, or when traffic volume is the primary objective. It does not optimise for what happens after the click.
6. Target impression share
Designed for visibility goals rather than conversion goals. You set a target percentage of auctions in which your ad should appear (top of page, absolute top, or anywhere on the page), and the algorithm bids accordingly. Useful for brand defence or competitor conquesting campaigns where presence matters more than direct response.
7. Enhanced CPC (eCPC)
A hybrid approach that adjusts your manual bids up or down based on the likelihood of conversion. Worth noting: Google has been deprecating eCPC for Search and Display campaigns, so this is not a long-term foundation. If you are currently relying on eCPC, plan a migration to a fully automated strategy.
Goal-to-strategy mapping at a glance:
| Business goal | Recommended strategy | Key metric to monitor | Minimum data requirement |
|---|---|---|---|
| Lower cost per sale or lead | Target CPA | CPA, conversion rate | 30–50 conversions per month |
| Higher revenue per £ spent | Target ROAS | ROAS, conversion value | 50+ conversions with value data |
| Maximum conversion volume | Maximise conversions | Conversion volume, CPA trend | Budget-constrained; no hard minimum |
| Maximum revenue volume | Maximise conversion value | Conversion value, ROAS trend | Value tracking required |
| Traffic and awareness | Maximise clicks | Clicks, CTR, CPC | None |
| Brand visibility | Target impression share | Impression share, average position | None |
For a deeper look at how these strategies compare in practice, the types of Google Ads bidding strategies guide covers selection criteria in detail.
How automated bidding actually works: signals, auctions, and learning
Every time someone searches, an auction runs. Manual bidding sets a static maximum bid for that keyword. Automated bidding does something fundamentally different: it calculates a unique bid for that specific user, in that specific context, at that specific moment.
Smart Bidding is Google’s label for its conversion-focused automated strategies (Target CPA, Target ROAS, Maximise conversions, Maximise conversion value). What separates Smart Bidding from basic automated strategies is auction-time bidding: the model does not set a daily average bid. It sets a precise bid for each individual auction, drawing on a wide range of contextual signals simultaneously.
The principal signals Smart Bidding uses at auction time:
- Device type: mobile, tablet, or desktop, each with different conversion rates
- Location: city, region, and proximity to physical stores
- Time of day and day of week: conversion probability shifts significantly across hours and days
- Remarketing list membership: whether the user has visited your site, abandoned a cart, or converted before
- Browser and operating system: Safari on iOS converts differently from Chrome on Android for many retailers
- Language settings: the user’s browser language preference
- Search query: the specific words used, not just the matched keyword
- Ad characteristics: the specific creative being shown
No human bidder can process eight signals simultaneously for thousands of auctions per hour. That is precisely why Google’s technical guidance notes that Smart Bidding outperforms manual bidding in dynamic environments: it captures signal combinations that are statistically significant for conversion likelihood but practically impossible to manage by hand.
Smart Bidding uses auction-time bidding and a wide range of contextual signals — including device, location, time of day, remarketing list membership, language, and operating system — to set a tailored bid for every single auction, optimising for conversions or conversion value rather than just clicks.
About Smart Bidding, Google Ads Help
How the algorithm learns
The model does not start from scratch with your account’s data alone. Smart Bidding uses query-level performance modelling, which means it trains on account-wide conversion data and borrows signal strength from related queries. For low-volume keywords with little individual history, the model draws on broader account patterns to make accurate bids. This is why accounts with mixed traffic volumes, some high-volume terms and some niche long-tails, often see the biggest gains from automation: the model pools conversions across query variants rather than waiting for each keyword to accumulate its own history.
A typical learning period runs 1–4 weeks depending on conversion volume. During this time, performance can look erratic. Avoid large structural changes (new campaigns, major budget shifts, landing page overhauls) while the model stabilises.
Pro Tip: Set a conversion window that matches your actual sales cycle before switching to any Smart Bidding strategy. If your customers typically take several days from click to purchase, a very short attribution window will starve the model of the data it needs.
When automated bidding is the right choice (and when it is not)
Automated bidding is not universally better than manual control. The right answer depends on your data quality, conversion volume, and campaign maturity.
Where automated bidding has a clear advantage:
- You have reliable conversion tracking and consistent conversion volume
- Your campaigns run across multiple devices, locations, and audience segments simultaneously
- You want to respond to real-time signal combinations that manual adjustments cannot capture
- You are scaling an ecommerce account where bid management at keyword level becomes operationally unsustainable
Where manual bidding or a hybrid approach still makes sense:
- New accounts with fewer than 30 conversions per month: the model lacks the data to optimise meaningfully
- Highly seasonal or promotional campaigns where historical data is a poor predictor of near-term behaviour
- Accounts with broken or incomplete conversion tracking: automation amplifies bad data, not just good data
- Situations where you need granular control over specific keywords for strategic reasons (brand terms, competitor terms)
Prerequisites before switching:
- Conversion tracking verified and firing correctly in Google Ads or Microsoft Advertising
- Conversion values assigned (for ROAS strategies)
- Attribution model reviewed and set appropriately (data-driven attribution is generally preferred for Smart Bidding)
- Budget set at a level that allows the strategy to learn: a budget cap that prevents the algorithm from spending freely will constrain learning
The hybrid approach worth considering: an agency manages strategy selection, conversion tracking quality, and structural decisions, while the platform’s algorithm handles bid-level execution. This is where specialist oversight adds measurable value, particularly for ecommerce brands running Shopping and Performance Max alongside Search.
Setting up automated bidding correctly: a practical checklist for UK advertisers
Getting the setup right shortens the learning period and prevents the most common performance problems. Work through these steps before switching any campaign to an automated strategy.
Essential setup steps:
- Verify conversion tracking end-to-end. Check that your Google Ads conversion tag or Google Tag Manager implementation fires on the actual confirmation page, not just the checkout initiation. For UK ecommerce retailers, test across mobile and desktop separately, as tag firing can differ by device.
- Set conversion values accurately. For Target ROAS to work, every conversion action needs a value. Use dynamic values pulled from your order confirmation page rather than static averages.
- Choose the right attribution model. Data-driven attribution is the default recommendation for Smart Bidding. Last-click attribution undervalues upper-funnel keywords and distorts the model’s understanding of what drives conversions.
- Exclude irrelevant conversion actions. If you track micro-conversions (newsletter sign-ups, page views) alongside purchases, make sure only the primary conversion action is included in the “Conversions” column used for bidding. Mixing them confuses the model.
- Size your budget appropriately. A rough rule: your daily budget should be at least 10–20 times your Target CPA to give the algorithm room to operate. A £10 daily budget with a £50 Target CPA is a structural constraint, not a bidding problem.
- Label your experiments. When testing a new strategy against an existing one, use Google Ads Experiments (formerly Drafts and Experiments) to run a proper A/B split. This gives you statistically meaningful data rather than a before/after comparison affected by seasonality.
UK-specific notes:
Currency settings should be confirmed as GBP before launch. Cross-border campaigns targeting both UK and EU audiences need separate conversion tracking considerations post-Brexit, particularly for VAT-inclusive versus VAT-exclusive revenue values. If your conversion value represents revenue including VAT, your Target ROAS target needs to account for that, or you will be optimising toward a gross figure that overstates true margin.
Pro Tip: During the learning period (typically 1–4 weeks), monitor the “Learning” status badge in your bid strategy report daily. If a campaign stays in learning longer than expected, the most common causes are insufficient conversion volume, frequent budget changes, or a target set too aggressively for current performance.
For practical tactics to reduce cost per conversion once your strategy is live, the guide on reducing cost per conversion in Google Ads covers optimisation steps that complement automated bidding well.
Common problems after switching to automated bidding and how to fix them
Most automated bidding problems trace back to one of four root causes: bad data going in, unrealistic targets, structural interference, or insufficient volume. Here is a diagnostic checklist.
Check these first:
- Is conversion tracking firing correctly? Pull a conversion action report for the past 30 days and compare conversion counts against your actual orders or CRM data. A significant discrepancy means the model is learning from inaccurate signals.
- Is the conversion window appropriate? A window shorter than your actual purchase cycle will under-report conversions and make the strategy appear to underperform.
- Are budget caps constraining the algorithm? If your campaign is hitting its daily budget cap regularly, the algorithm cannot bid freely. Either increase the budget or lower the target to match what the budget can realistically achieve.
- Has there been a recent structural change? Adding new ad groups, changing match types significantly, or restructuring campaigns resets the learning period. Performance volatility after a structural change is expected, not a sign the strategy is failing.
Quick fixes for rising CPA or falling ROAS:
- Raise your Target CPA or lower your Target ROAS by 10–15% to give the algorithm more room. Aggressive targets on thin data cause the model to under-bid and miss volume.
- Add negative audiences to exclude segments that consistently convert poorly (for example, users who have already purchased if you are not running a retention campaign).
- Shorten or lengthen your conversion window to better match actual purchase behaviour.
- Switch temporarily to Maximise conversions without a target if you need to rebuild conversion volume before reintroducing a constraint.
When to bring in specialist support:
Persistent CPA volatility over more than four weeks, despite clean tracking and reasonable targets, usually points to an account structure problem rather than a bidding problem. For Shopping and Performance Max campaigns, feed quality issues (missing attributes, incorrect categorisation, poor product titles) are a frequent hidden cause of poor automated bidding performance. If your Shopping ads are not converting profitably, the Shopping ads troubleshooting guide covers the most common feed and structure issues in detail.
Three practical scenarios: matching goals to the right strategy
1. High-volume ecommerce retailer
A UK fashion retailer running 200+ transactions per month across Google Shopping and Search has reliable revenue tracking and a clear margin target. The right strategy is Target ROAS, set at a level that reflects the margin after cost of goods. Setup priority: dynamic conversion values from the order confirmation page, data-driven attribution, and a budget at least 15 times the average order value per day. Monitor ROAS, conversion value, and impression share weekly. For scaling Shopping campaigns profitably alongside this, the Google Shopping strategy guide is directly relevant.
2. Lead generation with a longer sales cycle
A B2B software company generates a moderate number of qualified leads per month with a typical consideration period of about two weeks. Conversion volume is below the ideal threshold for Target CPA, so the better starting point is Maximise conversions without a target, running for 6–8 weeks to build data. Once a sufficient number of conversions accumulate, introduce a Target CPA set slightly above the current average to avoid over-constraining the model immediately. Assign conversion values based on estimated lead quality (for example, demo requests valued higher than content downloads) to prepare for a future move to Target ROAS.
3. Brand awareness and visibility campaign
A new product launch needs presence at the top of search results for branded and category terms. Target impression share is the appropriate strategy here, set to “Absolute top of page” for brand terms and “Top of page” for category terms. The trade-off: CPCs will be higher than a conversion-focused strategy because the algorithm bids for position rather than efficiency. Set a maximum CPC cap to prevent runaway spend on competitive terms, and monitor impression share and average CPC weekly rather than conversion metrics.
An agency perspective on automation and where human strategy still matters
There is a version of the automated bidding conversation that goes: “Set it up, let the algorithm run, and check back in a month.” That version is wrong, and it is responsible for a lot of wasted budget.
Automation handles bid-level execution better than any human can at scale. What it cannot do is decide which goal to optimise for, whether your conversion tracking is actually measuring the right thing, or whether your account structure is giving the algorithm a fair chance to learn. Those are strategic decisions, and they require human judgement.
At Oxedent, the ecommerce accounts that get the most from Smart Bidding are the ones where the strategic groundwork is solid before automation is switched on: clean conversion data, correctly valued transactions, sensible campaign structures, and a budget that matches the target. The algorithm then has something real to work with. Accounts that hand automation a broken tracking setup or an account with 15 overlapping campaigns tend to see the algorithm optimise confidently toward the wrong outcome.
The other underappreciated point: automated bidding is not a one-time setup. Targets need reviewing as margins change, seasonality requires proactive adjustments, and feed quality for Shopping and Performance Max needs ongoing attention. The specialist agency support guide covers why ecommerce brands with meaningful budgets tend to outperform those managing this in-house without dedicated PPC expertise.
Oxedent’s ecommerce PPC management: built around profitable automation
Automated bidding delivers its best results when the strategy, tracking, and account structure behind it are set up correctly from the start. That is exactly where Oxedent focuses.
Oxedent is a specialist ecommerce PPC agency working with established UK and US online retail brands. The agency manages Google Ads, Google Shopping, Performance Max, and Facebook Ads campaigns with a clear emphasis on ROAS and profitable revenue growth, not clicks or impressions. There are no long-term contracts, and every engagement starts with a clear audit of what your current setup is doing and where the gaps are.
Services directly relevant to automated bidding include Smart Bidding strategy selection and setup, conversion tracking audits, feed optimisation for Shopping and Performance Max, and ongoing campaign management with regular target reviews. If you are spending at least £2,000 per month on paid media and want a specialist team managing your automated bidding strategy, get in touch with Oxedent to discuss what a managed engagement looks like for your brand.
Sources
The following official platform pages are the primary references for automated bidding configuration and strategy selection:
- About automated bidding – Google Ads Help