Two of the three bidding changes Google just announced are optional. The third one isn't. Starting August 17, Google is changing how it optimizes campaigns limited by budget, and it applies to every account running Target CPA or Target ROAS. No toggle. No opt-in. That's the one worth your attention right now.
Google Ads Liaison Ginny Marvin shared the details on LinkedIn in mid-June 2026: Smart Bidding Exploration goes global, Promotion mode enters beta, and backend bidding target optimization gets reworked for budget-constrained campaigns. Account notifications start July 6 with historical performance data and recommendations. The calibration period that follows could cause temporary fluctuations in CPA and ROAS reporting.
Here's what each change actually means for demand gen operators running paid search, and what to do before August.
Smart Bidding Exploration: More Volume, Same Targets (Maybe)
Smart Bidding Exploration (SBE) is now available globally for Search campaigns and Performance Max campaigns without a product feed. A beta is open for Shopping and PMax with feeds. Google's pitch: SBE finds incremental converting traffic beyond your current query set without requiring you to loosen ROAS targets.
Google reported a 27% average increase in unique converting users for Search campaigns using SBE, per Search Engine Journal's coverage. That number sounds strong. The catch for B2B teams: "unique converting users" doesn't mean qualified pipeline. Without offline conversion data or value rules feeding back into Smart Bidding, the algorithm optimizes toward whatever conversion action you've defined. If that's a form fill, you'll get more form fills. Whether those form fills turn into revenue is a different question entirely.
Optmyzr's 2026 benchmark data underscores the tension. Year over year, CTR climbed from 1.83% to 2.22%, but conversion rate dropped 0.96% and CPA rose 4.41%. More clicks, fewer conversions per click, higher cost per conversion. The signal is clear: volume expansion without quality feedback loops makes the math worse, not better.
The trade-off with SBE: you might capture net-new converters, but if your conversion signal is weak (top-of-funnel only, no value differentiation), you're handing Google permission to spend on lower-intent queries. Test it, but only after your conversion setup reflects actual pipeline value.
Promotion Mode Beta: Useful, With Guardrails
Promotion mode lets you temporarily increase budget flexibility and relax ROAS tolerance for short-term events: product launches, webinar pushes, conference weeks, seasonal spikes. It works with daily budgets or campaign total budgets, and it's available in beta for Search and Performance Max.
For B2B SaaS teams that manually crank budgets and adjust targets around events, this could reduce operational overhead. The risk is discipline drift. A "temporary" ROAS relaxation that runs two weeks past the promo window becomes a permanent efficiency hit. Predefine the promo window, set a hard budget cap, and build a post-promo normalization plan before you flip it on. Document what "normal" looks like so you have a baseline to return to.
Beta availability isn't guaranteed for every account yet. Check before planning around it.
The August 17 Budget-Limited Update: The One That Matters Most
This is the change that applies whether you touch anything or not. Google is updating how it handles bidding target optimization for campaigns that regularly hit budget limits. Per Marvin's post: the goal is "more predictable performance in line with CPA and ROAS targets, especially when budgets increase."
Translation: Google wants to reduce the volatility spike that happens when a budget-constrained campaign suddenly gets more room. That volatility has been a real headache for teams managing monthly budget pacing. But the calibration period introduces its own uncertainty. Google hasn't specified how long calibration lasts or how large the fluctuations could be.
The operational risk here is compounding changes. If you adjust your CPA target and your budget and the backend optimization all shift within the same window, you lose the ability to isolate what caused performance movement. Separate your changes. Don't touch budgets and bid targets simultaneously during the calibration window.
Run It This Week
Before July 6: Pull a list of every campaign flagged as "Limited by budget" that uses Target CPA or Target ROAS. Document current targets, 30-day average CPA/ROAS, and conversion volume. That's your baseline.
July 6–August 17: Review Google's account notifications when they appear. Compare their historical data against your own records. Confirm your CPA and ROAS targets still reflect actual business goals (not targets set 6 months ago for a different pipeline stage).
For SBE testing: Only enable it on campaigns where you're feeding offline conversions or value-based bidding signals. Hypothesis: if we enable SBE on campaigns with offline conversion feedback, then unique converting users will increase without degrading SQL-to-opportunity rate, because the algorithm has a quality signal to optimize against. Success = 10%+ lift in unique converters with SQL rate within 5% of baseline. Guardrail: if CPA rises more than 15% over a 4-week window with no corresponding pipeline lift, pause the test.
For Promotion mode: Define the promo window, budget ceiling, and acceptable ROAS floor before activating. Run a pre/post comparison at 2 weeks post-promo to measure whether efficiency returned to baseline.
What to measure: Primary metric is cost per qualified opportunity, not cost per conversion. Secondary: conversion volume and SQL rate. Stop-loss: CPA exceeding 20% above target for 3+ consecutive weeks with flat or declining pipeline.
The broader pattern across all three updates is the same one that's defined paid search in 2026: Google keeps expanding automation's reach. The winning response isn't resistance. It's feeding the system better signals so the machine optimizes toward revenue, not just clicks. Teams that treat Smart Bidding as a black box will get black-box results. Teams that wire in offline conversions, enforce change-log discipline, and hold a 4-to-6-week evaluation window will actually know what's working.
The August 17 change doesn't care whether you're ready. July 6 is your heads-up. Use it.
