Lately, many performance marketers feel like glorified budget-adjusting machines. A common frustration in the industry is Google Ads seemingly ignoring target ROAS (tROAS) constraints, a runaway trend that is also becoming highly visible in Meta’s Advantage+ Shopping Campaigns (ASC). When you bring these concerns to platform representatives, the standard response is almost always a scripted: "Lower your tROAS to give the system more room to learn."

But does this advice actually serve your business, or does it just serve the platform's ad spend targets? To break out of this cycle, we need to understand the underlying algorithmic logic and implement advanced controls.

The Algorithmic Dilemma: High tROAS vs. Low tROAS

To understand why campaigns stall or bleed budget, we have to look at how the bidding engine behaves under different constraints:

  • When you set a high tROAS: The algorithm becomes hyper-conservative. It only bids on search queries and audiences with a near-guaranteed probability of conversion. Consequently, your impressions and traffic plummet.
  • When you lower your tROAS: You compromise to get traffic, and the system indeed gets the data it needs. However, it also starts bidding on a massive volume of low-intent, low-quality traffic to fill the volume gap.

This issue is compounded in multi-channel campaign types like Performance Max (PMax). Once you loosen ROAS constraints, the algorithm naturally shifts budget toward cheaper but lower-converting placements, such as the Google Display Network or YouTube. The result? Your front-end ad spend is fully utilized, but your back-end financial margins suffer. The AI lacks business context; it does not understand net profit, only the conversion metrics you tell it to optimize for.

Three Advanced Strategies to Reclaim Control

To protect your margins and scale efficiently in this automated era, you need to move away from basic bidding adjustments and implement structural changes.

1. Strategic Asset Orchestration

Avoid the trap of dumping your entire product catalog into a single, catch-all PMax campaign. When you do this, the algorithm will naturally favor a few top-selling SKUs to hit its targets, leaving the rest of your inventory starved of traffic.

Instead, segment your PMax campaigns based on product margins or sales velocity:

  • High-Margin Products: Group these together and apply a slightly lower, more aggressive tROAS to capture maximum volume and scale.
  • Low-Margin Products: Group these separately and apply a strict, high tROAS to protect your bottom line.

By physically separating your assets, you prevent the algorithm from averaging out performance across high- and low-performing products.

2. Leverage First-Party Data and Value Rules

An algorithm is only as good as the data it consumes. If you feed it raw, unadjusted conversion data, it will optimize for volume regardless of quality. To fix this, you must feed the system refined business data.

Implement offline conversion tracking (OCT) to pass back real-time data on product returns, cancellations, and actual net margins. Additionally, utilize Value Rules within Google Ads to adjust conversion values based on geographic location, device, or specific audience segments. Instead of fighting the bidding algorithm at the front end, you reshape its valuation of conversions at the back end, steering it toward high-value, profitable customers.

3. Build a Keyword Moat with Exact Match

By default, PMax will attempt to capture search traffic for your core brand terms and high-converting generic keywords if they are not actively targeted elsewhere. This often inflates PMax's reported ROAS by cannibalizing traffic that would have converted anyway through standard search.

To prevent this, ensure your core brand terms and high-intent conversion keywords are locked down in dedicated Search campaigns using Exact Match. When Google detects an exact match keyword in an active Search campaign, it will prioritize that campaign over PMax. This keeps your most valuable, high-intent traffic under tight, manual control, forcing PMax to do what it was designed to do: discover incremental, long-tail, and cross-channel audiences.

Conclusion

No matter how advanced Google and Meta's AI models become, they lack intrinsic business intuition. Winning in the era of black-box advertising requires performance marketers to retain control over traffic segmentation and data inputs. By managing how the algorithm is fed and where it is allowed to spend, you ensure that automation serves your net profitability, not just the ad platforms' bottom line.

Disclaimer: This article is based on public industry data and the author's professional analysis. It is intended for informational purposes only and does not constitute financial, investment, or legal advice.