Google Ads AI Bidding: Which Strategy to Use in 2026

Google Ads AI bidding has been around long enough that most advertisers have strong opinions about it — usually formed after at least one expensive mistake. tCPA blew the budget. Maximize Conversions sent CPCs through the roof. tROAS starved the campaign of volume. Sound familiar?

The honest truth is that Google's smart bidding algorithms are genuinely powerful in 2026. They process thousands of signals per auction — device, location, time, search query semantics, audience membership, browser, even weather patterns — faster than any human bid manager ever could. The problem isn't the technology. The problem is that most advertisers either apply the wrong strategy for their business objective, or they apply the right strategy on accounts that aren't ready for it.

This guide breaks down every major Google Ads AI bidding strategy, when to use each one, and the conditions your account needs to meet before automation can actually work in your favour.

Why Smart Bidding Is Table Stakes in 2026 — But Still Gets Misused

Google has been steadily removing manual control options. Enhanced CPC is effectively deprecated for most campaign types. Manual CPC still exists, but Google's own auction dynamics — with auction-time quality signals that manual bidding can't access — mean you're fighting with one hand tied behind your back if you rely on it exclusively.

That said, "set a smart bidding strategy and leave it" remains one of the most common and costly mistakes we see in accounts at Workflow AI Advisors' paid media practice. Smart bidding requires clean conversion tracking, adequate data volume, realistic targets, and regular strategic oversight. Without those foundations, Google's algorithm optimises toward the wrong thing — or optimises aggressively toward the right thing but in ways that destroy margin.

Let's go through each strategy properly.

Maximize Conversions (Without a Target CPA)

What it does: Google spends your entire daily budget to get as many conversions as possible, regardless of what each conversion costs.

When to use it: This is a launch strategy, not a scaling strategy. Use it when a campaign has fewer than 30 conversions in the past 30 days and you need to generate enough signal for Google's algorithm to learn. It's also appropriate when budget is genuinely uncapped and volume is the only objective — rare, but it happens in lead gen for high-value B2B pipelines where even 5 leads per month justifies significant spend.

The risk: Without a tCPA constraint, Google will find conversions at whatever price the auction requires. In competitive verticals, CPCs can climb significantly during the learning phase. We typically set a portfolio budget cap and monitor CPA daily for the first two weeks.

Account readiness requirement: At least one well-configured conversion action with accurate tracking. No minimum conversion volume required — that's the point of using this strategy first.

Target CPA (tCPA)

What it does: Google optimises bids to get the maximum number of conversions at or below your target cost per acquisition. It will bid aggressively in auctions where it predicts a conversion is likely, and pull back in auctions where it doesn't.

When to use it: This is the right strategy for lead generation, app installs, and any business where the value of each conversion is roughly equal. If you're generating enquiries for a service business and every qualified lead is worth approximately the same to you, tCPA is your default.

The critical constraint: Google officially recommends a minimum of 30–50 conversions per month per campaign. In practice, we've seen tCPA perform well with as few as 20–25 monthly conversions in stable, low-competition verticals. Below that threshold, the algorithm doesn't have enough signal and will either over-restrict or over-bid erratically.

The most common mistake: Setting a target CPA that's too aggressive relative to historical performance. If your account has been converting at £85 CPA on manual bidding, don't immediately set a £55 tCPA target. Start at £90–£95, let the algorithm stabilise over 2–3 weeks, then pull the target down gradually — no more than 10–15% at a time.

At Workflow AI Advisors, our standard tCPA optimisation workflow — which includes structured target adjustments, audience layering, and negative keyword governance — has delivered an average 31% CPA reduction across client accounts without sacrificing conversion volume.

Target ROAS (tROAS)

What it does: Google optimises for conversion value rather than conversion volume. It bids to achieve a target return on ad spend — for example, a 400% tROAS means Google will try to generate £4 of conversion value for every £1 spent.

When to use it: tROAS is built for ecommerce, where different products have different margins and different average order values. It's also appropriate for any business that passes dynamic conversion values into Google Ads — lead scoring models, LTV-weighted lead values, or tiered service values.

The data requirement is higher: Google recommends 50 conversions with values in the past 30 days. We'd push that recommendation up to 70–100 for tROAS specifically, because value optimisation is more complex than volume optimisation. Thin data produces volatile bidding behaviour.

The margin trap: A common error is setting tROAS based on revenue rather than gross margin. If your blended gross margin is 45% and you set a 200% tROAS (2x revenue return), you're losing money on every sale. Your tROAS target must be calculated backward from margin, not from revenue alone. This sounds obvious — it's missed constantly.

Performance Max and tROAS: In 2026, the vast majority of ecommerce accounts running Performance Max campaigns are using tROAS as the bidding backbone. The interaction between PMax asset groups, tROAS signals, and product feed quality has become complex enough that it warrants its own dedicated strategy review — something we cover in detail in our paid media service.

Maximize Conversion Value (Without a Target ROAS)

What it does: The value-equivalent of Maximize Conversions. Google spends your full budget to maximise total conversion value, without a ROAS constraint.

When to use it: Use this as a launch strategy for ecommerce campaigns before you have enough conversion value data to support tROAS. It's also useful during aggressive scaling phases where you want Google to find high-value customers and you're willing to accept variable ROAS while volume builds.

The nuance: This strategy often produces impressive headline numbers — high total conversion value — while quietly delivering mediocre ROAS. Always monitor both volume and efficiency metrics simultaneously.

Target Impression Share

What it does: Bids to achieve a specific share of impressions — either anywhere on the results page, at the top, or at the absolute top. Google adjusts bids to win the impression share percentage you specify.

When to use it: Brand campaigns. Competitor conquesting campaigns where visibility is the primary goal. Situations where brand safety or market presence genuinely outweighs direct response efficiency — launching in a new market, for instance.

When not to use it: For any performance-driven campaign. Target Impression Share will happily spend enormous budgets chasing impression share in auctions with zero conversion intent. We see this misapplied to non-brand search campaigns regularly. It is not a performance strategy.

Enhanced CPC (eCPC)

What it does: Adjusts your manual CPC bids up or down based on Google's conversion likelihood signals. A hybrid between manual and automated bidding.

Current status in 2026: Google has deprecated eCPC for Search campaigns and is actively deprecating it for other campaign types. If you're still using it, plan your migration to a fully automated strategy. eCPC is not a long-term option.

Choosing the Right Strategy: A Decision Framework

Rather than a rigid flowchart, here's how we think about strategy selection in practice:

Step 1 — Verify conversion tracking. Before anything else, audit your conversion actions. Are you tracking the right events? Are values being passed correctly? Is there deduplication between Google Ads and GA4? Broken conversion tracking feeding into smart bidding is worse than no automation at all — the algorithm will optimise toward phantom conversions or misattributed signals.

Step 2 — Assess conversion volume. Less than 30 monthly conversions per campaign? Start with Maximize Conversions. Build volume first. Once you hit 30–50 conversions consistently, migrate to tCPA with a conservative target.

Step 3 — Define your primary objective. Volume (leads, installs, form fills)? tCPA. Value efficiency (ecommerce revenue, margin-weighted leads)? tROAS once you have the data. Awareness or market presence? Target Impression Share, but only for brand or strategic campaigns.

Step 4 — Set targets based on data, not aspiration. Pull your last 60–90 days of performance. Set your initial tCPA or tROAS target at or slightly above the historical average. Optimise from there incrementally.

Step 5 — Give the algorithm time, but not unlimited time. Google's learning period is nominally 1–2 weeks. In practice, allow 3–4 weeks before drawing conclusions after any significant change. But if a campaign is materially underperforming after 6 weeks with no improvement trend, don't wait for magic — reassess the strategy, the target, or the underlying campaign structure.

Signals Smart Bidding Uses (That Manual Bidding Can't)

Understanding what Google's algorithm actually sees helps calibrate expectations appropriately. In 2026, auction-time bidding signals include: user's real-time search query and intent signals, device type and operating system, geographic location down to city level, time of day and day of week, the user's recent search history and browsing patterns, audience list memberships (remarketing, Customer Match, Similar Audiences), the specific ad and landing page being served, competitive auction dynamics in real time, and contextual signals including weather in some verticals.

No human bid manager can process these simultaneously per auction. This is why smart bidding, when applied correctly, outperforms manual bidding in most mature accounts. The key phrase is "when applied correctly."

What AI Bidding Can't Fix

Smart bidding amplifies what already exists in your account. It cannot compensate for weak ad creative, poor landing page experience, a mis-targeted keyword strategy, or a broken conversion funnel. If your Quality Scores are low and your landing pages convert at 1.2%, smart bidding will spend your budget efficiently reaching users who still won't convert.

This is why our approach at Workflow AI Advisors pairs bidding strategy with the full account infrastructure — keyword architecture, audience strategy, creative testing, and organic visibility working in alignment. Paid media doesn't perform in isolation.

2026 Account Structures That Work With Smart Bidding

The old "one keyword per ad group" (SKAG) structure actively fights against smart bidding by fragmenting conversion data. In 2026, consolidated campaign structures — broader ad groups with strong audience and creative signals — give smart bidding algorithms the data density they need to perform. This doesn't mean abandoning match type discipline or negative keyword hygiene. Those remain critical. It means building structures that pool enough signal for the algorithm to learn effectively.

Performance Max has further complicated this by absorbing Search, Shopping, Display, YouTube, and Gmail inventory into unified campaigns. The bidding strategy within PMax is always automated — the lever you control is the quality of your asset groups, audience signals, and conversion data. Getting that right is where the real work happens.

If your current account structure was built for a manual bidding world, it likely needs rebuilding before smart bidding will perform at its potential. Our AI automation service includes account structure audits that identify exactly where campaign architecture is limiting algorithmic performance.