The problem: the account works, but nobody can explain why
Google Ads management becomes a different job once an account passes its first year. The structure has been patched by three different people. There are 14 conversion actions and nobody knows which ones feed bidding. Performance Max is reporting conversions that brand search used to claim. Spend has grown, the blended ROAS still looks fine, and yet finance is asking why new-customer growth has flattened.
At this stage the improvements do not come from adding keywords or writing new headlines. They come from decisions about architecture, data quality and automation control that most agencies never revisit after the initial build. That is the work this page describes. If you are earlier in the journey, or need cross-platform paid search, start with our PPC management service.
What we mean by Google Ads management
Managing a Google Ads account well means owning six recurring decisions: how the account is structured, how queries are matched, how the platform’s bidding models are fed and constrained, how Performance Max is given assets and signals, which conversion data the account learns from, and how budget is scaled without breaking any of the above. Everything else, from ad copy tests to extensions, sits on top of those six.
Who this service is for
- Accounts spending enough per month that a 10% efficiency gain is worth more than the management fee, and where the current setup has stopped improving.
- Retailers relying heavily on Performance Max and Shopping who suspect the campaign is cannibalising brand and existing-customer sales.
- B2B and SaaS teams whose form fills do not equal pipeline and who need the account to bid on lead quality rather than lead volume.
- In-house marketers who want a specialist to run the account, or an audit and restructure they can then operate themselves.
What Axoria does
Google Ads account audit
A structured review of settings, structure, match types, search terms, conversion actions, bidding, Performance Max configuration, asset quality, audience signals, change history and wasted spend. Delivered as a prioritised document with estimated impact per fix, and available as a standalone project.
Account structure decisions
Whether to consolidate or split campaigns, where brand belongs, how to separate margins, geographies and customer types, and how many campaigns the account’s conversion volume can actually support. Documented so the logic survives staff changes.
Match type and query control
A deliberate policy on broad, phrase and exact match by campaign, paired with Smart Bidding where broad is used and with negative lists and brand exclusions everywhere. Search-term hygiene reviews on a fixed cadence.
Quality Score improvement
Ad relevance, expected CTR and landing page experience diagnosed at keyword level. Low scores on high-spend terms are fixed by tightening ad groups, rewriting responsive search ads and improving landing page relevance and speed.
Smart Bidding strategy
Choosing between Target CPA, Target ROAS, Maximise Conversions, Maximise Conversion Value and manual approaches based on conversion volume, value accuracy and business goals, and managing targets through learning periods and seasonality.
Performance Max management
Asset group design by product or theme, audience signals, listing group structure, brand exclusions, feed quality in Merchant Center, and the use of campaign-level negatives and search themes to keep spend where it belongs.
Conversion actions and enhanced conversions
A clean conversion set with one primary goal per campaign type, values assigned, enhanced conversions for web and for leads configured, and consent mode handled correctly.
Offline conversion import
Lead status and deal value fed back from your CRM (HubSpot, Salesforce or via sheets and Zapier) so the account bids towards qualified pipeline and closed revenue, not form submissions.
How we make account decisions
Structure follows data volume
Smart Bidding learns per campaign. An account with 60 conversions a month split across 15 campaigns gives each bidding model four conversions to learn from, which is not enough. We consolidate until each campaign has a meaningful signal, then split only where a segment genuinely needs its own budget or target, such as a product line with half the margin of the rest. The reverse is also true: a single campaign hiding a loss-making region is split out.
Match types are a control setting, not a keyword setting
Broad match with Smart Bidding can find profitable queries you would never have listed, and it can also spend heavily on irrelevant ones. We use it in campaigns with strong conversion signals and tight negatives, keep exact and phrase for high-value terms where message control matters, and never run broad match with manual bidding. Each campaign’s match policy is written down.
When to use Smart Bidding, and when not to
Target CPA and Target ROAS need a steady flow of accurately valued conversions. Below roughly 30 conversions a month per campaign, or when conversion values are placeholders, the models make poor decisions. In those cases we use Maximise Conversions with a bid cap through a portfolio strategy, or feed a higher-volume proxy conversion (a qualified pricing page visit, for instance) while offline import builds up. Once the data supports it, we move to value-based bidding and adjust targets gradually rather than in large jumps that restart learning.
Performance Max: give it good inputs and boundaries
Performance Max performs in proportion to the quality of what you give it. For a retailer with 3,000 SKUs, that means product titles in the feed that lead with the attributes people search for, complete GTINs and product types, images that meet the spec, and asset groups organised by category so the creative matches the products. We exclude brand terms so the campaign cannot claim sales that were already coming, set new-customer acquisition goals where the business wants growth rather than retention, and review the placement and search-term insight reports monthly.
Conversion data is the real account structure
Everything above depends on the account learning from the right outcomes. We reduce conversion actions to a small primary set, mark the rest as secondary for observation, assign values that reflect real business worth, enable enhanced conversions to recover matches lost to browser restrictions, and connect offline conversion import for lead-based businesses. Consent mode is configured so that modelled conversions are available in regions where consent rates are low.
Example: bidding on pipeline instead of forms
A B2B software account generating 200 form fills a month, of which 40 become sales-qualified, is bidding on the wrong thing if it optimises to form fills. Importing the “SQL” status from the CRM with a value, and setting that as the primary conversion, lets Target CPA seek the auctions that produce qualified leads. Reported lead volume usually falls; qualified volume typically rises. That trade is worth explaining to stakeholders before it happens.
Scaling and budget pacing
Scaling a Google Ads account is not the same as raising the budget. Increasing daily budgets by more than about 20% at a time or changing targets sharply can push Smart Bidding back into learning and inflate CPAs. We scale in steps, watching impression share lost to budget, search impression share on priority terms, and marginal CPA per campaign. Where a campaign is already winning most of its available auctions, extra budget goes to new query territory, new geographies or a different campaign type rather than the same one.
Budget pacing is managed to the month with shared budgets or scripts where useful, so the account does not overspend in week one and go dark in week four. Seasonality adjustments are applied ahead of known peaks and removed afterwards.
Our Google Ads process
Full account audit
Two to three weeks depending on account size. Structure, tracking, bidding, Performance Max, search terms, assets, change history and wasted spend, with a prioritised fix list and estimated impact.
Conversion foundation
Rebuild conversion actions, implement enhanced conversions and consent mode, and connect offline conversion import. This step happens before any restructure so the new structure learns from clean data.
Restructure
New campaign architecture built in Google Ads Editor, reviewed with you, and migrated in phases so historical performance is not lost overnight.
Bidding and Performance Max tuning
Strategies set per campaign with explicit rules for when targets change. Asset groups, feed improvements and exclusions applied to Performance Max.
Optimise and scale
Search-term reviews, asset tests, target adjustments and budget steps on a documented cadence, with a monthly report on what changed and why.
Tools we typically work with
Google Ads, Google Ads Editor and Google Ads scripts for build, bulk changes and pacing. Google Merchant Center and a feed management tool for Shopping and Performance Max. GA4 and Google Tag Manager for measurement, with server-side tagging where consent and browser restrictions justify it. HubSpot, Salesforce or Zapier for offline conversion import. Looker Studio for reporting. Semrush or Auction Insights for competitive context. Landing page issues are handled with our CRO team, using Microsoft Clarity or Hotjar.
How results are measured
| Question | Metric we use | Source |
|---|---|---|
| Is the account producing qualified outcomes? | Cost per qualified lead or profit on ad spend, by campaign | Google Ads with offline import; margin data from you |
| Is growth coming from new customers? | New-customer share of conversions, especially in Performance Max | Google Ads new-customer reporting, CRM |
| Is automation behaving? | Learning-period days, CPA variance week to week, target vs actual | Google Ads bid strategy reports |
| Are we buying incremental sales? | Geo or conversion-lift tests on brand and Performance Max | Google Ads experiments, internal analysis |
| Where is spend being wasted? | Spend on non-converting search terms and placements, impression share lost to rank on priority terms | Search-term and placement reports |
The tracking stack behind these reports is built through our analytics and tracking service, and where paid social runs alongside search, we report both in one view so budget can move between them on evidence rather than habit. For background on how these pieces fit together, our PPC guide and performance marketing guide cover the fundamentals.
What to expect
Reported numbers may drop before they improve
Excluding brand from Performance Max, removing duplicate conversion actions and bidding on qualified leads all reduce the conversions the account reports. They also make the account honest. We show the before-and-after logic so the change is understood internally.
Restructures take time to settle
Migrating an account resets some learning. We phase changes over several weeks and expect four to eight weeks before the new structure is fairly judged.
We will not follow every recommendation
Google’s automated recommendations and auto-apply settings are turned off. Each suggestion is evaluated against your goals, and many are declined.
No guarantees on ROAS or CPA
Auction dynamics, offers and landing pages shape results. We commit to a documented plan, honest reporting and early warning when targets look wrong.