Increasing ROAS on Paid — From Bleeding Budget to Profitable Acquisition

The Problem

The client's paid programme had worked well. For two years, Google Ads had been a reliable, efficient source of pipeline with consistent ROAS that justified significant and increasing investment. Then, over about eight months, performance began to deteriorate. Not dramatically — never a sudden cliff — but a persistent, gradual decline that continued despite multiple attempts to address it. CPC was up 35% from the previous year. Conversion rate was down 22%. ROAS had fallen from a healthy 4.1x to a loss-making 1.8x.

The team had tried the obvious fixes. They'd refreshed ad creative. They'd adjusted bids. They'd tried new audiences. They'd moved budget between campaigns. Nothing worked for more than two to three weeks before performance slipped back toward the deteriorating trend. By the time we were engaged, the paid programme was generating pipeline at a cost that made it economically irrational to continue at current spend levels, but with no clear understanding of why it had stopped working or what to do about it.

The Diagnostic Audit

We ran a comprehensive audit across every dimension of the paid programme. The findings revealed not one problem but five simultaneous issues that had compounded to create the overall decline. Any one of them alone might have been manageable. All five together created the sustained, difficult-to-fix performance collapse the client was experiencing.

Issue one was audience saturation. The core targeting had not meaningfully changed for eighteen months. The same audience was being shown the same types of creative over and over. When we pulled historical data on impression frequency for the core audience segments, the numbers were alarming — core ICP audience members were seeing ads from this brand an average of 22 times per month. Fatigue was severe. Click-through rates had fallen 45% from their peak, and the users still clicking were increasingly marginal prospects rather than high-quality leads.

Issue two was broad match creep. Over the previous twelve months, Google's Smart Bidding system had gradually expanded match types, and the account had not been maintained with the negative keyword discipline required to control this. When we pulled the search terms report for the previous ninety days, over 30% of spend was going to queries that no reasonable account manager would have approved if reviewed manually — irrelevant industries, consumer rather than B2B intent, informational queries with no commercial application.

Issue three was landing page decay. The primary conversion pages had not been updated in fourteen months. What had been a strong, differentiating value proposition when the pages were built was now generic — competitors had caught up on messaging, the offer was no longer distinctive, and the social proof featured companies that the target audience no longer found impressive. The pages looked dated and felt underpowered relative to what well-funded competitors were offering.

Issue four was attribution drift. A website migration eight months earlier had introduced subtle conversion tracking issues. Free trial signups were being tracked correctly, but demo requests — a more valuable conversion type with higher downstream pipeline value — had a broken tracking tag. Smart Bidding had been optimising toward the wrong conversion type for eight months, not because anyone made a bad decision, but because a technical change broke a critical signal without anyone noticing.

Issue five was competitive cost inflation. Simply put, three significant new competitors had entered the market in the previous eighteen months, all with substantial funding and aggressive bidding strategies. CPCs for core keywords had risen significantly across the market. The client's bids hadn't adequately adapted to the new competitive reality.

The Recovery Plan

We addressed the issues in priority order based on estimated impact and implementation complexity. The attribution fix came first — this was urgent because Smart Bidding was making decisions based on incorrect data, and every day it continued was compounding the misalignment. The tracking issue was resolved within a week, and we rebuilt the conversion configuration to correctly represent the full value of each conversion type.

The negative keyword audit was second. We paused broad match keywords entirely for the first month and ran phrase and exact match only while building a comprehensive negative keyword list from the search terms data. This immediately reduced wasted spend by approximately 28% and improved the relevance signal that Smart Bidding was receiving.

The creative overhaul ran over six weeks — developing four entirely new creative directions, each testing a different angle and value proposition, with fresh visual treatments that bore no resemblance to the fatigued creative that had been running for eighteen months. The new creative was tested systematically with modest budget before scaling the winners.

The landing pages were rebuilt from scratch on every page receiving significant paid traffic. New headline testing, new social proof featuring recent, recognisable customers, new offer structure, and page-level message matching to ensure the specific promise made in each ad creative was echoed immediately on the page the user arrived at.

The audience strategy was expanded to break out of the saturated core segment and begin building reach into lookalike and intent-based segments while the core audience was rested and re-warmed.

The Outcome

Three months after beginning the recovery programme, ROAS had improved from 1.8x to 3.2x — a meaningful improvement in programme economics that made scaling investment rational again. CPL from the highest-quality segments fell 40% from the trough as negative keyword management dramatically reduced wasted spend. Lead quality scores from the sales team improved, reflecting better targeting and message-market fit from the new creative and landing pages.

The broader lesson from this engagement is that paid programme decline is rarely caused by a single fixable issue. It's almost always a compounding of multiple factors — audience saturation, technical issues, creative fatigue, competitive dynamics, landing page quality — that need to be diagnosed and addressed simultaneously. Fixing one factor at a time while others continue compounding produces disappointingly modest results. A comprehensive reset across all underperforming dimensions is typically required to meaningfully reverse a sustained decline.