From 3x to 6x ROAS: a lead gen case study in the office industry

A ROAS of 3x often feels like a respectable result. You're earning three euros back for every euro invested in Google Ads, and that's better than breaking even. But in the office industry, where average deal values are high and the sales cycle runs longer, 3x is rarely the ceiling. It's usually just the starting point. In this case study, we walk through how a phased, AI-driven approach brought a lead gen campaign in the office industry from 3x to more than 6x ROAS — in four quarters, with no increase in budget.

The starting point: a solid foundation, but growth had stalled

The campaign we took over wasn't poorly built. The account had a clear structure, ads were equipped with relevant extensions, and the advertiser had some experience with Smart Bidding. Yet ROAS had been stuck at around 3.1x for three months. Every attempt to improve performance through manual bid adjustments or new ad copy produced temporary spikes, but no structural growth.

This pattern is common in service-oriented sectors, including the office industry. Google Ads is working, but without the right data input, the right signals, and a continuous optimization cycle, the system plateaus. The campaign runs at a local optimum while the real potential is significantly higher.

Analysis revealed three core issues. First, conversion signals were incomplete: tracking was implemented client-side, meaning a portion of actual lead submissions wasn't making it back into Google Ads. Second, search term analysis was done manually once a week, allowing irrelevant terms to burn budget for days before being added as negative keywords. Third, the Responsive Search Ads (RSAs) had sat unchanged for months, with Ad Strength scores of "Average" or worse.

Phase 1: strengthening conversion signals with server-side tracking

ROAS development per quarter

Office industry lead gen campaign — before and after AI optimization (2026)

Q1 — Baseline (manual)

3.1x

Q2 — After structure optimization

4.2x

Q3 — After Smart Bidding switch

5.1x

Q4 — Fully AI-driven

6.3x

The first step was fixing the tracking problem. Using our own server-side GTM infrastructure (sGTM), all conversion moments were rebuilt with Enhanced Conversions and first-party data enrichment. This means Google Ads now receives richer signals about who actually submits a quote request: hashed email address, geographic data, and device type are all sent with the conversion event.

The direct effect was immediate: within two weeks of implementation, 28% more conversions were being tracked than before. These weren't new leads — they were leads that had always existed but were previously going unmeasured. This directly influences Smart Bidding, because the algorithm receives more data to learn from. Stronger signals lead to better bidding decisions, which in turn drive down the cost per lead (CPL) and push ROAS higher.

For lead gen campaigns, this effect is even more significant than in e-commerce, because conversion values aren't always immediately visible in the pixel. By assigning a modeled value to each quote submission — based on historical close rate and average contract value — we were able to activate Target ROAS as a bidding strategy instead of the simpler Maximize Conversions. This gives the system a clearer goal and makes optimization considerably sharper.

Phase 2: daily search term mining and negative keywords

Alongside the tracking improvements, automated search term mining began. Where previously a search terms report was reviewed manually each week, our AI now analyzes all incoming search terms daily. Irrelevant terms are detected immediately and added as negative keywords, often within 24 hours of first appearing.

In the first month of automated mining, more than 140 unique search terms were added as negative keywords that would have burned budget for at least a week under manual management. Examples include terms like "free office furniture", "second-hand office chairs", or informational queries such as "how much does an office renovation cost on average" — terms that generate clicks but rarely lead to serious quote requests.

The freed-up budget was redistributed toward the best-performing search terms, which were first safely tested via our Keyword Incubator in a separate incubator campaign before graduating to the production campaign. This ensures that only proven converting keywords receive the full budget allocation.

• Average Quality Score rose from 6.2 to 7.8 across all active keywords
• Average CPC dropped by 18% due to improved ad relevance and search term match
• Percentage of irrelevant clicks fell from approximately 22% to less than 6%
• Total lead output increased by 31% with no budget increase
• ROAS reached 4.2x in Q2, a 35% improvement over the starting point

Phase 3: RSA optimization and improving ad performance

Lead gen conversion funnel office industry

Average flow after AI optimization (2026 benchmarks)

Impressions

Targeted impressions via Search and PMax

Clicks

Average CTR 7% after RSA optimization

Lead page visited

38% of clicking visitors reach the quote form page

Quote submitted

18% conversion rate on the landing page

Cost-efficient leads via AI-driven Smart Bidding

In the third phase, the AI optimization focused on the ad copy itself. The existing RSAs had mostly an Ad Strength of "Average" or even "Poor" at audit. This signals that Google doesn't consider the combinations of headlines and descriptions optimal for the targeted search queries.

Our RSA improvement system automatically analyzes Ad Strength scores per ad group and generates new headlines and descriptions based on the best-performing combinations in the sector, supplemented with signals from the landing page. Ads with a "Poor" status are automatically rewritten and submitted for approval before going live.

Within six weeks, all RSAs in the campaign were upgraded to at least "Good", with a significant portion reaching "Excellent". This directly impacts Ad Rank and therefore the position of the ad in the auction. A higher Ad Rank at equal or lower CPC is the ideal scenario for ROAS improvement.

Alongside the RSA improvements, the audience strategy was also automated. Through our audience management system, RLSA audiences were automatically created and refreshed weekly: visitors to the quote form page who hadn't converted received an increased bid, while already-converted visitors were excluded from the campaign. This further reduces waste and increases the quality of the audience reached.

How AdBrains AI handles this structurally

What this case study illustrates is that doubling ROAS isn't the result of one magic intervention, but of a set of continuously running optimization processes working in concert. That's exactly what AdBrains' proprietary AI technology is built for.

The foundation is our multi-agent verification system. Every optimization decision — whether it's a bid adjustment, a negative keyword, or an RSA change — is reviewed by four independent AI agents before execution. This eliminates errors that occur in manual management or single-agent AI systems. An agent that wants to add a search term as a negative keyword is flagged if another agent signals that the same term has historically converted in another campaign. This prevents conflicts that, in manual management, lead to serious budget waste.

On top of this verification system runs automatic tCPA/tROAS optimization: bidding strategies per campaign are adjusted daily based on current conversion volume and the margin targets set per client. When a campaign temporarily drops below the minimum conversion threshold, our strategy-switch system automatically shifts to a safer bidding strategy to prevent quality loss, and reactivates the original strategy once data flow recovers.

The Keyword Incubator plays a crucial role in safe growth. New keywords detected via search term mining as potentially promising are first placed in a controlled incubator campaign with a limited budget. Only once they've built sufficient conversion data do they automatically migrate to the production campaign, protecting the main campaign's performance at all times.

Finally, our server-side signal enrichment ensures that the conversion signals Google receives are always as rich and complete as possible. Via our own sGTM infrastructure, first-party data points such as CRM signals, lead quality scores, and customer lifetime values are fed back into Google Ads as Enhanced Conversions. This gives Smart Bidding the most detailed possible picture of which clicks generate truly valuable customers, directly benefiting Target ROAS optimization.

Results: from 3x to 6x in four quarters

Manual management vs. AdBrains AI

Campaign performance comparison office industry lead gen

Manual management

• ROAS: 3.1x (stagnation after 3 months)
• Weekly manual search term review
• Static bidding strategy (Maximize Clicks)
• RSAs rarely updated
• Negative keywords added reactively
• No audience segmentation
• Basic conversion signals (client-side)
• Campaign expansion: monthly, manual

AdBrains AIRecommended

• ROAS: 6.3x (growth every quarter)
• Daily automated search term mining
• Dynamic Target ROAS via Smart Bidding
• RSA improvement system (auto-fix POOR status)
• Negative keywords proactively added daily
• Automated RLSA and audience management
• Server-side signal enrichment (Enhanced Conversions)
• Auto campaign expansion based on growth data

The combined impact of all optimization phases is visible in the quarterly figures. Starting at a ROAS of 3.1x in Q1, the campaign grew to 4.2x in Q2 following the tracking and search term improvements. In Q3, the Smart Bidding switch combined with RSA improvements pushed ROAS to 5.1x. By the end of Q4, with the fully automated system running at full capacity, the campaign reached a ROAS of 6.3x.

To put this in a lead gen context: for every euro of ad spend, 6.30 euros of pipeline value was being generated by the end of the process. With a constant monthly budget, that translated to more than twice as many qualified quote requests — at no extra cost. CPL fell by 41% over the same period, while average lead quality (measured via CRM data and close rates) demonstrably improved.

This pattern is consistent across other lead gen clients such as Clima-Active.nl, which offers air conditioning and heat pump installation. In the HVAC sector, just like in the office industry, quote requests are the primary conversion objective and deal values are high enough to make Target ROAS a viable bidding strategy. The phased AI approach consistently delivers stronger results in this type of campaign than manual management or generic automation.

The following table summarizes the most impactful differences between manual management and the AdBrains AI system in the context of lead gen:

The table makes clear that the ROAS doubling isn't coincidental: it's the logical result of dozens of small improvements applied daily, while manual management consistently falls short on each of these dimensions.

What manual management structurally misses

The contrast between manual campaign management and AI-driven optimization is especially sharp in the office industry. The reasons are multiple:

• Slow reaction time: manual optimization happens weekly or bi-weekly. In a real-time Google Ads auction, that's too slow to respond to shifts in search behavior or competitive pressure.
• Incomplete data analysis: a human account manager cannot analyze thousands of search terms, hundreds of ad combinations, and dozens of audience segments every day. AI does this continuously.
• Suboptimal bidding: manual bid management leads to adjustments based on intuition or limited datasets, rather than on all available conversion data.
• Reactive, not proactive: negative keywords are added manually after budget has already been wasted; our system detects and blocks irrelevant terms daily before they consume meaningful spend.
• No scaling efficiency: expanding a campaign manually requires additional time and resources; with AdBrains AI, optimization scales automatically as the campaign grows.

Frequently asked questions about ROAS optimization in lead gen

How long does it take to grow from 3x to 6x ROAS?

In this case study, the trajectory took four quarters — one full year. The speed of growth depends on three factors: the quality of the existing account structure, the available conversion volume (more conversions means faster-learning Smart Bidding algorithms), and the level of server-side tracking implementation. For campaigns that already have a strong foundation, significant ROAS improvements are sometimes visible after just two quarters. The first month is typically the most impactful, as tracking improvements and negative keyword mining produce immediate effects.

Is Target ROAS suitable as a bidding strategy for all lead gen campaigns?

Target ROAS works best when sufficient conversion data is available (guideline: at least 30-50 conversions per month per campaign) and when each conversion has been assigned a value. For lead gen campaigns, this means defining a modeled value per quote submission based on your historical close rate and average contract value. Without this value assignment, Target Conversions or Maximize Conversions is a better starting point. AdBrains AI helps define the right conversion values and automatically switches between bidding strategies based on available conversion volume.

What are the most common pitfalls in lead gen ROAS optimization?

The biggest pitfall is optimizing for leads rather than lead quality. A low CPL sounds attractive, but if leads don't convert into customers, the true ROAS is much lower than the measured ROAS suggests. This is why feeding CRM data back into Google Ads is essential, so that the Smart Bidding algorithm learns to optimize for leads that actually become clients. A second common mistake is raising Target ROAS targets too quickly, causing the system to operate in too narrow a search volume and lose impressions. AdBrains AI monitors this threshold automatically.

How does the Keyword Incubator work in practice?

The Keyword Incubator is a separate campaign within the Google Ads account that serves as a testing environment for new keywords. When our search term mining detects a term that appears potentially promising but has no proven conversion history, that term is added as an exact match or phrase match keyword to the incubator campaign with a limited budget. Once it has built sufficient conversion data and its CPA or ROAS performance exceeds the defined threshold, the keyword automatically migrates to the production campaign. This protects the main campaign's performance while enabling safe, data-driven growth of the keyword portfolio.