Book an intro meeting →
Google AdsPaid Media

Google Ads Smart Bidding in 2026: A Practical Guide

Using Google Ads Smart Bidding in 2026: choosing between tCPA and tROAS, setting realistic targets, surviving the learning period and feeding it better data.

Google Ads Smart Bidding in 2026: A Practical Guide

Most advertisers have an opinion about Smart Bidding, and plenty of those opinions were formed years ago after a campaign went sideways. The product has changed a lot since then. In 2026, automated bidding is the default for almost every campaign type in Google Ads, and the question is no longer whether to use it but how to set it up so it works for your business rather than against it.

Here's how we approach Smart Bidding: which strategy to pick, how to set targets, what actually happens during the learning period, and the data that makes the difference between a campaign that scales and one that quietly burns budget.

How Smart Bidding actually works

Smart Bidding uses Google's machine learning to set a bid for every individual auction. Instead of one bid per keyword, it looks at the signals attached to each search: device, location, time of day, the exact query, browser, audience lists, past behaviour and many more. Crucially, it weighs those signals in combination, which is something no manual bid adjustment can do.

That's its real advantage. A manual bidder can raise bids for mobile, or for Brisbane, or for weekday mornings. Smart Bidding can recognise that a returning visitor on a laptop in Brisbane searching a specific product at 9pm is far more likely to convert than the average click, and bid accordingly.

"Smart Bidding is only ever as smart as the conversion data you give it."

Choosing the right strategy

Google's bid strategies now come down to two families, each with an optional target:

Without a target, the strategy simply tries to get the most conversions (or value) it can for your budget. That's useful for a new campaign that needs to gather data. Adding a target tells Google what efficiency you need, and it will hold back spend on auctions that don't meet it.

A simple rule of thumb: if you can't say with confidence that a sale or lead from one campaign is worth more than another, start with target CPA. If you can pass real values, such as order revenue or a lead's likely worth, value-based bidding will almost always serve the business better.

Setting targets you can live with

The most common mistake we see is setting an aggressive target on day one. Ask for a $40 CPA on a campaign that has historically delivered leads at $70, and Google will simply stop entering most auctions. Spend drops, volume disappears, and the conclusion is usually "Smart Bidding doesn't work".

The learning period, explained

Whenever you launch or significantly change a bid strategy, it enters a learning period while it recalibrates. Google says this can take up to around 50 conversion events or three conversion cycles, depending on how much data is available. A conversion cycle is how long it typically takes someone to convert after clicking, which could be minutes for an online order or weeks for a high-value service.

A new learning period is triggered when:

  1. A bid strategy is created or switched back on
  2. Any bid strategy setting is changed, including the target
  3. Campaigns, ad groups or keywords are added to or removed from a portfolio strategy

Two things follow from this. First, every tweak resets the clock, so batching changes is far better than making a small adjustment every few days. Second, low-volume campaigns take much longer to learn. If a campaign only produces a handful of conversions a month, consider combining similar campaigns or using a portfolio strategy so the algorithm has more data to work with.

Don't judge week to week. Conversions are often reported days after the click. Look at performance over a rolling 30 days, and use the conversion estimates in the bid strategy report, which allow for conversion delay, rather than raw recent numbers.

Feed the algorithm better data

This is where most of the real gains are. Smart Bidding optimises towards whatever you tell it a conversion is, so the quality of that signal matters more than any setting.

1. Bid on the right conversion

Only your primary conversion actions are used for bidding. Make sure they're the outcomes that matter: a purchase, a qualified lead or a booked call. Page views, scroll depth and newsletter sign-ups belong as secondary actions. Pick the lowest-funnel action that still has enough volume. If true sales are too rare, a qualified-lead event is usually a better target than a raw form fill.

2. Turn on enhanced conversions

Enhanced conversions send hashed first-party data, such as an email address from a checkout or enquiry form, so Google can match more conversions back to ad clicks. With cookie restrictions and browser privacy changes, it's one of the simplest ways to stop the algorithm flying blind.

3. Import offline conversions

If your sales happen off the website, in a CRM, over the phone or in person, import those results back into Google Ads. A lead-gen campaign optimising towards form fills will happily find you cheap, low-quality enquiries. One optimising towards leads that became customers will find you better ones.

4. Pass real values

Value-based bidding needs values that reflect the business. For eCommerce, that's revenue at a minimum and ideally profit margin. For lead generation, assign values by lead type or service line. If every conversion carries the same value, target ROAS has nothing meaningful to optimise towards.

Tools worth knowing about

Common Smart Bidding mistakes

  1. Changing too much, too often. Every change restarts learning. Make deliberate changes, then leave them alone.
  2. Budget as the bottleneck. If your budget runs out by lunchtime, Smart Bidding can't pick the best auctions across the day. The target should be what limits spend, not the budget.
  3. Optimising for the wrong conversion. Counting phone-number clicks or page views as primary conversions teaches Google to find people who click, not people who buy.
  4. Judging too early. A bad first week during learning is normal. Decide on a rolling 30-day view.
  5. Ignoring search terms. Automation doesn't replace housekeeping. Review search terms and add negatives regularly.

When manual bidding still makes sense

There are still a few situations where we'd hold off: a brand-new account with no conversion tracking yet, campaigns with so little volume they'll never exit learning, and very short promotions where historical data is misleading. In those cases, Maximise clicks or manual CPC can bridge the gap while you build data. For everything else, a properly set-up Smart Bidding strategy should outperform manual bidding over any reasonable time frame.

The bottom line

Smart Bidding isn't a set-and-forget button, and it isn't a black box either. Pick the strategy that matches how your business makes money, set realistic targets, give it time to learn, and above all, feed it conversion data that reflects real outcomes. Get those right and the algorithm becomes your best bid manager.

Running into the same mistakes over and over? Read the most common paid ads mistakes, or if you'd like a second set of eyes on your account, talk to us about Google Ads management.

Need a hand with this?
Google Ads with Base Digital

Search, Shopping and Performance Max campaigns built around profit, with tracking you can trust.

Explore Google Ads
Free strategy call
Want a second opinion on your marketing?

Book a free call with Brandon. Honest advice on where your biggest opportunities are, with no hard sell.

Book a call
Keep reading

More from the Knowledge Base

View all articles →
Google Ads, Paid Media22 October 2026
How to Audit Your Google Ads Account in an Hour
Read article →
Meta Ads, Paid Media22 October 2026
Meta Ads Tracking After iOS: What Still Works
Read article →
eCommerce, Paid Media6 October 2026
Black Friday 2026: The Sales Tactics That Actually Make You Money
Read article →