Google Smart Bidding · How tCPA, tROAS, and Friends Actually Work
Google's automated bidding strategies — Target CPA, Target ROAS, Maximize Conversions, Maximize Conversion Value — explained without the marketing fluff. What each does, when to use it, what data they need, and the failure modes to watch for.
Attribution. Smart Bidding strategies are documented in Google Ads Help and discussed extensively by Google Ads product team members publicly. This article synthesizes public documentation with operator experience from RGM client accounts.
What Smart Bidding actually is
Smart Bidding is Google's family of automated bid strategies that use machine learning to set per-auction bids in real time based on signals beyond what a human bidder could process. The signals include device, location, time, audience, browser, search query specifics, and many others.
The trade-off: you give up granular bid control in exchange for the algorithm using more signals than you can. For most accounts at meaningful scale, this trade is favorable. For very small accounts or highly bespoke bidding strategies, manual bidding can still win.
The main strategies
Strategy
Goal
When to use
Target CPA (tCPA)
Hit a specific cost-per-acquisition
Lead gen, fixed-value conversions
Target ROAS (tROAS)
Hit a specific return on ad spend
E-commerce with variable order values
Maximize Conversions
Get the most conversions for the budget
Brand awareness, lead volume goals without strict CPA
Maximize Conversion Value
Get the most revenue for the budget
E-commerce, value-tracked conversions
Maximize Clicks
Get the most clicks for the budget
Traffic goals, brand visibility
What Smart Bidding needs to work
Three things are non-negotiable:
Conversion volume. Smart Bidding needs ~30 conversions per campaign per month to learn well. Below that, the algorithm doesn't have enough signal to outperform manual bidding.
Accurate conversion tracking. Garbage in, garbage out. If conversions are double-counted, mis-attributed, or missing, the bidding optimizes against the wrong signal.
Sufficient learning time. 1–2 weeks of stable conditions before judging performance. Changing campaigns mid-learning resets the algorithm.
The most common Smart Bidding failure we see. Teams switch to Target CPA, see immediate cost-per-conversion fluctuations, panic, and switch back to manual after 3 days. The algorithm needed 10–14 days to stabilize. The fix: commit to a 14-day learning period before judging, and don't change targets, budgets, or structure during it.
tROAS · the gotchas
tROAS optimizes for value, which means it needs conversion values set correctly. For e-commerce, this is usually dynamic (the actual transaction value). For lead gen with fixed conversion values, tROAS becomes essentially equivalent to tCPA.
tROAS bids more aggressively for high-value conversions and less for low-value ones. If your value tracking is wrong (e.g., reporting gross instead of net, including refunded transactions), tROAS will bid against wrong signals and reduce overall performance.
Smart Bidding with offline conversion import (OCI)
For long sales cycles where the value isn't known at the moment of click, offline conversion import (OCI) feeds CRM-confirmed conversions back to Google Ads, weighted by deal value when known. Smart Bidding then optimizes for closed-won, not just lead, dramatically improving lead quality.
OCI is the highest-leverage Google Ads optimization for most B2B advertisers and remains under-implemented.
Enhanced Conversions
Enhanced Conversions improves conversion accuracy by sending hashed first-party data (email, phone) alongside conversions. This helps recover attribution that ITP, ATT, and cookie deprecation otherwise break. Pair with Smart Bidding for materially better signal quality.