Account Hygiene To Signal Hygiene: The New Maintenance Tasks For PPC Teams

Paid search maintenance has changed. Google’s newer campaign systems push more matching, bidding, asset generation, and landing-page decisions into automation. Performance Max uses Google AI to optimize bids and placements and explicitly relies on advertiser inputs like customer data, creative assets, and conversion values. AI Max for Search can expand matching, optimize ad content, and use final URL expansion to send traffic to dynamically selected landing pages.

That shifts the job of the PPC team. The old model centered on manual bid tuning, constant keyword sculpting, and micro-adjustments. The new leverage is signal hygiene: making sure the conversion data, values, landing pages, creative assets, audience inputs, and exclusions feeding the system are clean enough that the algorithms learn the right lesson. In other words, modern optimization is less about steering every auction and more about preventing the machine from optimizing toward noise.

What You’ll Learn Today

• PPC maintenance now depends more on input quality than on constant manual tuning, especially in AI-led formats like Performance Max and AI Max for Search.

• The highest-leverage maintenance task is conversion hygiene: choosing the right primary conversions, keeping low-quality actions secondary, and making values reflect real business value.

• Negative keywords, brand exclusions, and campaign segmentation still matter, but now they are mainly there to prevent bad learning, not to micromanage every query.

• Landing-page governance is now a signal-quality issue because AI Max can use final URL expansion and dynamic landing pages. Bad routing teaches the system to send traffic to the wrong destinations.

• Asset hygiene matters because Google can automatically create or customize assets, and ads, assets, and destinations still go through policy review.

• A practical PPC team should run a recurring weekly and monthly signal-hygiene checklist instead of relying on reactive optimization alone.

Why maintenance changed

The core change is structural. Google Ads now automates more of the levers PPC teams used to manage directly. Performance Max optimizes bids and placements with Google AI, while AI Max for Search can influence search-term matching, ad content, and URL expansion. That means the inputs you give the system now matter more than many of the old manual adjustments.

That does not mean maintenance matters less. It means the maintenance work moved upstream. Instead of asking, “What bid tweak should we make today?” the better question is, “What bad signals are we feeding the model right now?” That is the difference between account hygiene and signal hygiene.

Account hygiene vs signal hygiene

Account hygiene was the old model. It focused on surface-level account management: keyword list cleanup, bid modifiers, ad rotation choices, placement exclusions, and constant manual tuning.

Signal hygiene is the new model. It focuses on what the system learns from:

• which conversions are used for bidding

• how those conversions are valued

• whether tracking is accurate and deduplicated

• whether landing pages match the intent and the ad promise

• whether assets are compliant and role-appropriate

• whether exclusions are blocking useless traffic before it corrupts the learning loop

The point is not that old PPC tasks disappeared. It is that they now serve a different purpose.

1. Conversion data hygiene is the highest-leverage task

Google Ads lets teams classify conversion actions as primary or secondary. Primary actions are used for bidding optimization and appear in core conversion reporting. Secondary actions are not directly used for optimization in the same way and are better treated as diagnostic or supporting signals.

That distinction is now foundational. If the system optimizes toward cheap but low-quality events, the campaign can look efficient while actually training itself on the wrong outcome. This is especially dangerous in lead generation, where form fills, brochure downloads, or other easy actions can drown out true qualified pipeline signals.

A clean conversion hierarchy usually means:

• primary conversions are tied to revenue, qualified pipeline, or another real business outcome

• secondary conversions are micro-actions used for diagnosis, not bidding

• campaign-specific goals are only used when truly needed, because Google says account-level goals usually let campaigns learn from one another more effectively, and Smart Bidding needs time to adapt when conversion settings change

Value hygiene matters just as much. Google’s conversion value rules can be used in real time by Smart Bidding for Target ROAS and Maximize Conversion Value, and Google explicitly positions them as a way to reflect business goals like revenue, profit, offline value, or lifetime value more accurately.

That means PPC teams should stop asking only, “Did this convert?” and start asking, “Was this the kind of conversion the algorithm should chase?”

2. Query and traffic hygiene prevents bad learning

Negative keywords still matter. Google defines them simply: they prevent your ad from being triggered by certain words or phrases. Brand exclusions also still matter, especially as Search brand settings move into AI Max workflows. For Performance Max and Search, brand exclusions are there to stop campaigns from serving on branded queries you want to avoid.

The strategic shift is why you use them. The goal is not obsessive control of every query. The goal is to block irrelevant, low-intent, or misleading themes before they generate junk conversions that teach the system the wrong pattern.

In practice, query hygiene means reviewing:

• irrelevant informational themes

• bargain or free-intent traffic when that is not your buyer

• competitor or brand ambiguity that does not match the campaign objective

• coupon, student, repair, job-seeker, or support intent bleeding into acquisition campaigns

For Performance Max, it also means reviewing placement waste and brand-safety issues where applicable, not because placement management is the core lever, but because obvious waste still weakens the account’s learning quality.

3. Landing-page and routing hygiene matters more than teams think

AI Max for Search can use final URL expansion and dynamic landing pages. Google’s own help documentation warns that dynamic landing pages can break with incompatible tracking templates and lead to 404 errors or other routing problems if the setup is not compatible.

This makes landing-page governance a signal issue. If automation expands into the wrong pages, the campaign can start collecting lower-intent traffic, weaker conversion quality, or policy risk from destinations that were never meant to be part of the campaign.

The fix is operational:

• maintain a small approved landing-page set for high-risk or regulated categories

• audit routing for dynamic URL expansion

• check whether the landing page actually matches the promise of the ad and the intent of the query

• treat wrong-page routing as a training-data problem, not just a UX problem

4. Asset hygiene is now part of risk management

Google can automatically create assets and text customizations when it predicts they will improve performance. Ads and assets also go through review, and Google reviews headlines, descriptions, keywords, destinations, images, and video. Content that violates policy can be disapproved or blocked.

That means PPC teams need an approved asset library, not just a pile of ad text. The system needs:

• claims-safe headlines and descriptions

• approved qualifiers and disclaimers

• clear proof snippets for regulated or high-scrutiny categories

• asset removal rules for poor performance or policy risk

Pinning should be used selectively, mainly when compliance or clarity requires it. Default pinning everywhere just limits the model without solving the underlying governance problem.

5. Audience and first-party signal hygiene still matters

Performance Max explicitly calls out customer data and audience signals as important advertiser inputs. Customer Match also only allows first-party data collected directly from customers, and Google requires consent where applicable. In the EEA, consent signals matter for ad personalization workflows. Consent mode lets tags adjust behavior based on user consent, and Google states that it helps improve measurement accuracy in a privacy-first way.

So audience hygiene is not just “upload a list.” It means:

• keeping customer lists refreshed

• segmenting by lifecycle stage and value, not just “all users”

• confirming first-party data eligibility

• making sure consent setup is aligned with how the account measures and targets

If those inputs are stale or non-compliant, the model is learning from the wrong audience picture.

6. Experiment hygiene matters because Smart Bidding has memory

Google says Smart Bidding models take time to adapt when conversion goals change. That means testing too many things at once, or changing goals and targets too abruptly, creates noisy readouts and unstable learning.

A better operating rule is simple:

• test one hypothesis at a time

• keep budgets stable enough for the model to learn

• separate structural changes from measurement changes when possible

• use incrementality or lift checks when feasible, especially if the account may be optimizing toward easy but low-value conversions

This is less exciting than rapid-fire optimization, but it produces better learning.

Practical examples

In B2B lead gen, the classic failure mode is optimizing to low-quality form fills because they are frequent and cheap. The signal-hygiene fix is to move qualified lead or qualified pipeline actions into the primary layer, keep weaker form events secondary, clean out spam or accidental submissions, and separate campaigns by intent stage so the system is not mixing evaluation traffic with real buying traffic.

In ecommerce, the common failure is letting Performance Max learn on low-margin SKUs or low-value coupon traffic. The fix is to calibrate values, use value rules where appropriate, segment campaigns or feeds by economics, and use exclusions to keep obviously low-quality traffic from dominating the learning loop.

In healthcare or other regulated categories, the risk is often in creative and landing-page generation. Automatically created assets or text customization can create phrasing that is too broad, and wrong-page routing can expose unsupported claims. The fix is an approved claims library, tighter landing-page governance, and explicit review gates before new assets or destinations are allowed to scale.

Weekly and monthly maintenance cadence

Weekly, about 60 to 90 minutes:

• review search-term waste and add negatives or exclusions where needed

• check for conversion anomalies such as spikes, duplicate patterns, or unexplained drops

• review landing-page mismatch and routing issues

• review assets for policy risk, weak performance, or off-message phrasing

Monthly, about 2 to 4 hours:

• audit primary vs secondary conversion actions

• recalibrate values where economics have changed

• run tracking QA for deduplication, cross-domain behavior, and consent-mode impact

• review campaign segmentation by intent, lifecycle stage, or margin

• run a learning review to see whether the system is optimizing toward the outcomes the business actually wants

That is the new maintenance model. The account still needs cleanup, but the highest-return work is no longer endless tweaking. It is protecting the quality of what the machine learns from.

Potenture’s Signal Hygiene Audit is built around that reality: audit conversion actions and values, tracking integrity, query and placement waste, asset compliance, and landing-page routing, then turn the findings into a recurring maintenance SOP that improves performance by fixing what the algorithms learn from.