A Google Ads agency can look busy while its portfolio quietly leaks money. That was the central problem in this agency portfolio optimization case study: too many accounts, too many alerts, and too little clarity about which issues were actually hurting client results.
The agency did not need another dashboard. It did not need an automation tool making silent decisions inside client accounts. It needed an independent way to identify waste, rank opportunities by financial impact, and give account managers a defensible action plan.
This is a composite case study based on common patterns in multi-account Google Ads portfolios. The numbers are illustrative, but the operational problems are painfully real.
The agency managed 38 Google Ads accounts across ecommerce, local services, B2B lead generation, and subscription businesses. Monthly media spend ranged from $4,000 to more than $180,000 per account. Every account manager had a routine: review performance, adjust bids, check search terms, respond to client questions, and build reports.
The routine was not the issue. The issue was that routine optimization favored what was visible over what was valuable.
A rising cost per conversion got attention. A poor-performing campaign got attention. But hidden waste often did not. Search terms with weak relevance stayed active. Broad match traffic was not consistently controlled. Quality Score problems were found late, if at all. PMAX results were accepted at face value because the underlying signals were hard to inspect.
Across 38 accounts, even a small blind spot became expensive. A 3% to 5% waste rate is not a rounding error when an agency manages millions in annual spend.
The agency's leadership set a simple goal: create a portfolio-wide process that could identify the highest-value fixes without forcing teams to rebuild campaigns or abandon their existing workflow.
The agency stopped treating every account as equally urgent. Instead, it reviewed accounts through two lenses: potential financial impact and confidence that a corrective action would improve performance.
That distinction mattered. A large account with a possible issue is not always a higher priority than a smaller account with a clear, recurring source of wasted spend. Scale matters, but evidence matters more.
The team began with a structured diagnostic across the full portfolio. It looked for four recurring categories of loss:
Rather than creating a 100-item checklist for every client, the agency assigned each finding an estimated value. The calculation was straightforward: affected spend, likely efficiency loss, and the level of control the team had to correct it.
A search-term issue affecting $20,000 in monthly spend with a clear negative-keyword solution rose to the top. A theoretical improvement with unclear attribution did not. This gave account managers permission to ignore low-impact busywork.
The first portfolio review identified opportunities in 31 of the 38 accounts. That did not mean 31 accounts were badly managed. It meant that normal management had not exposed every source of inefficiency.
The most common problem was search-term control. Several accounts had expanded into queries that looked relevant at a glance but did not match the client's commercial intent. A B2B software account, for example, was attracting research-heavy searches from users looking for templates, salary data, and free alternatives. Conversion volume was not catastrophic, but the spend had no credible path to pipeline.
The second major issue was fragmented Quality Score improvement work. Teams knew that ad relevance and landing-page alignment mattered. What they lacked was a fast way to identify where poor Quality Score was materially increasing cost. Some low-score keywords had little volume and deserved no attention. Others were high-spend terms that had become accepted as expensive simply because they had always been expensive.
The third finding was reporting complacency around automated campaign types. In several accounts, PMAX and broad match campaigns were producing conversions, but the agency could not confidently separate incremental growth from conversion capture. Brand demand, returning visitors, and existing customer activity were sometimes being credited as campaign success.
That did not mean automated campaigns were wrong. It meant they required scrutiny. Automation is not a strategy. It is a delivery mechanism that still needs commercial guardrails.
The agency divided findings into 30-day, 60-day, and monitor-only actions. This prevented the usual optimization mistake: making too many changes at once and losing the ability to explain what improved performance.
In the first 30 days, account managers focused on high-confidence waste reduction. They added negative keywords, tightened match-type controls where search terms showed weak intent, paused clearly inefficient targets, and corrected conversion actions that were inflating results. For lead generation clients, they also separated form fills from qualified leads wherever CRM feedback was available.
The next phase addressed Quality Score and ad relevance. Teams rewrote ads around high-value query themes, improved asset alignment, and flagged landing-page mismatches for client-side action. The agency did not promise that every Quality Score increase would lower CPA immediately. That would be nonsense. But it treated high-spend, low-score terms as a commercial problem worth testing, not a cosmetic account metric.
For PMAX, the agency used a more cautious approach. It reviewed campaign results alongside branded search behavior, audience signals, product margins, and new-customer trends. When data was incomplete, the instruction was not "turn it off." It was "stop assuming." Some campaigns earned more budget. Others had targets tightened, exclusions reviewed, or their role in the account reduced.
This is where an independent diagnostic layer such as Dolnai can help. It does not take over account management or alter campaign structures. It surfaces issues that are easy to miss when teams are buried in platform reporting and client delivery.
After 90 days, the agency measured results against a pre-optimization baseline and reviewed each account separately before reporting portfolio totals.
Twenty-four accounts reduced avoidable spend without a meaningful decline in conversion volume. Across those accounts, the agency estimated a 9% reduction in nonproductive search spend. CPA improved by more than 10% in 17 accounts, although the degree of improvement varied by vertical, tracking quality, and sales cycle.
The stronger outcome was operational. Account managers spent less time hunting for problems and more time making decisions. Weekly reviews became shorter because teams arrived with a ranked list of issues, not a blank screen and a vague instruction to optimize.
Client conversations improved as well. Instead of saying, "We made several adjustments," the agency could explain the commercial logic: this query group consumed budget without qualified leads, this keyword cluster had a relevance problem, and this campaign needed stricter proof before receiving more spend.
That level of specificity matters when a client asks why performance changed. It also protects agencies from the trap of reporting activity instead of accountability.
Portfolio optimization is not a universal recipe. Negative keywords can reduce waste, but overly aggressive exclusions can cut off valuable discovery traffic. Quality Score work can lower costs, but it cannot fix a weak offer or a landing page that fails to convert. PMAX may be overcredited in one account and genuinely incremental in another.
The point is not to apply the same fix everywhere. The point is to make uncertainty visible, then prioritize actions based on spend, evidence, and business impact.
Agencies also need clean conversion definitions. If an account counts page views, unqualified calls, and real sales leads as equivalent conversions, no optimization process can produce trustworthy conclusions. Bad measurement turns smart bidding into expensive guesswork.
The agency's lasting change was not a one-time cleanup. It created a monthly portfolio review with a simple discipline: identify the highest-cost inefficiencies, assign an owner, record the expected impact, and check whether the fix changed outcomes.
Each account manager still owned strategy. That is essential. A diagnostic system should sharpen judgment, not replace it. The best agencies do not hand control to black-box automation and call it scale. They build a repeatable way to see what others miss.
Your next portfolio review does not need more slides or more platform noise. Start with the spend that has the weakest business case, ask what evidence supports it, and make the clearest correction first.
‍
.svg)
