A campaign beating its target CPA by 50% sounds like a win. Until you realize the algorithm was bidding you into cheaper, lower-competition auctions to keep your spend within budget. That's the dynamic Google's Aug. 17, 2026 Smart Bidding update appears to have disrupted, and the early data paints a picture B2B SaaS teams need to sit with before adjusting anything.

What Google actually changed

Google updated how target-based Smart Bidding (tCPA, tROAS) behaves for campaigns flagged "Limited by budget." The stated goal: make performance trend more consistently toward the advertiser's stated target, regardless of whether the campaign is budget-constrained. The update rolled out starting Aug. 17 and completed Aug. 27, affecting Search, Shopping, Performance Max, and Demand Gen campaigns.

Google's framing, as reported via Ginny Marvin: campaigns that are both budget-limited and outperforming their target may now move closer to that target. Practitioners have a less diplomatic name for the prior behavior: "bid suppression." The hypothesis is that Smart Bidding was lowering bids in budget-limited campaigns to stretch spend across cheaper auctions. The result looked like outperformance, but it may have been the algorithm picking easier fights on your behalf.

The numbers so far (directional, not definitive)

Smarter Ecommerce, led by Mike Ryan, published an early account-sample analysis:

The pattern is consistent: budget-limited campaigns started competing more aggressively in auctions (higher bids, less rank-based loss), then hit their budget ceilings sooner (more budget-based loss, lower total impression share). The cheap-auction discount evaporated.

One caveat Ryan himself flags: conversion attribution takes time, and some advertisers changed their targets around the update. Treat these as leading indicators, not settled benchmarks.

Why this matters for B2B SaaS pipeline planning

If your account had budget-limited tCPA or tROAS campaigns that consistently "beat target," that outperformance may have been partially artificial. The algorithm was achieving it by competing in weaker auctions: cheaper clicks, not necessarily better ones. The update forces those campaigns toward your stated target unless you change the target itself. For growth leaders forecasting pipeline from paid search, your CAC assumptions may need recalibrating. A campaign that "always beat target" wasn't a signal of channel efficiency; it was a side effect of how Smart Bidding managed your budget constraint.

The diagnostic before you touch anything

  1. Identify exposed campaigns. Filter for tCPA or tROAS campaigns flagged "Limited by budget" that were consistently outperforming their target pre-Aug. 17.
  2. Segment impression share data. Compare IS lost to rank vs. IS lost to budget, pre and post Aug. 17. A shift from rank-dominated loss to budget-dominated loss confirms the pattern Smarter Ecommerce observed.
  3. Watch CPC and pacing. If median CPC jumped and the campaign is spending out earlier in the day or week, the bid suppression discount is gone in your account.
  4. Wait on ROAS/CPA conclusions. Conversion lag means the efficiency picture won't stabilize for several weeks. Don't panic-adjust targets based on two weeks of CPC data.

Google's own guidance (as reported) suggests a specific alternative for fixed-budget advertisers: switch to Maximize Conversions or Maximize Conversion Value without a target. The trade-off you're accepting: you lose the CPA/ROAS guardrail. For SaaS teams that must cap spend but want volume, this is worth testing, not adopting wholesale.

The experiment to run

The hypothesis (make it falsifiable): if we switch one budget-limited tCPA campaign to Maximize Conversions (no target) while holding budget constant, then cost per qualified lead at the SQL stage will stay within 15% of baseline because the algorithm will optimize for volume within the same budget, and we'll gain back impression share lost to budget.

Setup: Pick one campaign with clear pre/post data. Run for three weeks minimum (conversion lag). Primary metric: cost per SQL. Guardrails: CPA at the MQL stage stays within 25% of the 30-day pre-update average. Stop-loss: if CPA exceeds 30% of baseline after week two, pause and revert.

What to measure (and what not to over-interpret): CPC changes alone don't tell you whether pipeline quality shifted. You need to follow conversions through to qualified stages. Platform-reported ROAS is directional attribution, not proof of incrementality.

Where this leaves the auction

The cheaper auction inventory that budget-limited campaigns used to absorb? It appears to be flowing toward advertisers without budget constraints, which explains the 13% CPC drop in that cohort. If you're running uncapped campaigns alongside capped ones in the same account, check whether your uncapped campaigns saw CPC improvements over the same period.

Google described this as a predictability improvement. Practitioners see something more structural: the economics of budget-constrained advertising changed. Both framings can be true simultaneously. What matters for your planning is simpler. The targets you set are now the targets the algorithm pursues. If your target was aspirational rather than reflective of your actual marginal CAC/LTV, this update just made that gap visible.