A mid-market B2B SaaS client came to us in Q4 2025 spending $28,000 per month across Google Ads and LinkedIn, with a blended cost per lead of $312 and a sales team that described roughly 60% of inbound leads as "not worth calling." The core issue was not creative or targeting. It was that every channel was being evaluated on last-click, so budget kept flowing toward assets that looked efficient in isolation but were cannibalising each other in practice. Over 90 days, by rebuilding the attribution model and restructuring channel roles, we brought CPL down to $143 and increased sales-qualified lead volume by 38%.
The Starting Point: What Last-Click Was Hiding
The client's Google Ads account was returning a reported CPL of $89 on branded and competitor terms, which made it look like the most efficient channel by far. LinkedIn Sponsored Content was showing a CPL of $380 and was under pressure to justify its budget. What last-click attribution masked was that 71% of the Google Ads conversions came from users who had clicked a LinkedIn ad or read an organic blog post in the two weeks prior. Google was getting credit for closing demand that LinkedIn and content had built.
We pulled 6 months of CRM data and matched it against UTM paths and GA4 session sequences. The analysis showed that deals sourced from LinkedIn had a 34% higher close rate and a 22% higher average contract value than those attributed to Google Ads alone. The channel that looked most expensive was actually producing the best pipeline quality. Without that data, the instinct would have been to cut LinkedIn, which would have quietly collapsed the Google Ads performance within a quarter.
This pattern is more common than most teams realise. For a deeper look at why platform-reported numbers mislead B2B advertisers, our breakdown of multi-touch attribution for B2B ROI covers the methodology in detail.
Rebuilding Channel Roles Before Touching Budgets
Before we moved a single dollar, we redefined what each channel was supposed to do. LinkedIn became the primary demand-generation layer, targeting job titles (VP of Operations, Head of Finance) at companies between 200 and 2,000 employees. Its goal was awareness and intent signals, not direct form fills. Google Search was restructured to capture only high-intent, non-branded queries from users already in an active buying cycle, with tightly controlled match types and a negative keyword list that eliminated informational searches.
We also separated the Google Ads campaigns by funnel stage rather than by product line. Retargeting campaigns targeted users who had visited pricing or demo pages within 14 days, with messaging specific to objection handling rather than feature lists. Cold search campaigns used a much tighter keyword set, focused on terms indicating vendor comparison or immediate need. This separation made it possible to read performance at each stage without one segment contaminating another's data.
- LinkedIn: demand generation and account-level intent signals, optimised for video views and content engagement, not CPL
- Google Search (cold): 18 exact and phrase-match keywords targeting active-evaluation queries, no broad match
- Google Search (retargeting): separate campaign for visitors of /pricing and /demo pages, with bid adjustments up to +65%
- Offline conversion import: CRM deal stage data pushed back into Google Ads weekly to train Smart Bidding on qualified leads, not all form fills
The Attribution Fix: Data-Driven Over Last-Click
We switched the Google Ads account from last-click to Google's data-driven attribution model, which distributes conversion credit across touchpoints based on observed contribution patterns in the account's own data. For accounts with at least 300 conversions per month in the relevant conversion action, this model consistently produces more accurate bid signals than position-based alternatives. The client's account qualified comfortably, with around 480 tracked form fills per month.
The more impactful change was implementing offline conversion tracking tied to CRM opportunity stage. Previously, every form fill was treated as equal in the bidding signal. After the fix, only leads that reached "SQL" stage in HubSpot were imported as primary conversion actions. Form fills were downgraded to a secondary, non-bid-optimised action. Within three weeks, Smart Bidding began shifting spend away from keywords that generated high fill volume but low SQL rates, and toward the smaller set of terms that reliably produced qualified pipeline.
We also layered in a simple cross-channel tracking document updated weekly: a spreadsheet pulling LinkedIn campaign data, Google Ads data, and HubSpot pipeline data into a single view with a manually calculated blended CPL and a cost-per-SQL by channel. It is not sophisticated, but it forced every channel decision to be made against the same metric rather than each platform's native numbers.
What Changed in the Google Ads Account Specifically
On the Google Ads side, the structural work ran parallel to the attribution changes. We removed 214 keywords that had generated clicks but zero SQLs in six months, consolidated 11 ad groups into 4 tightly themed ones, and paused all broad match modifiers in cold campaigns. The account had been running Performance Max alongside standard search, and PMax was consuming 41% of budget while contributing 9% of SQLs. We reduced PMax to a single asset group targeting existing customer lookalikes only, with a strict budget cap.
Landing pages were another lever. The client had been sending all paid traffic to the product homepage, which carried six different CTAs and three separate navigation bars. We built two dedicated landing pages: one for the cold search campaign (focused on a single problem statement and a demo CTA) and one for retargeting (featuring a case study quote, a pricing anchor, and a 15-minute call offer). For context on why homepage routing consistently underperforms in B2B paid search, the patterns we see are well documented in our post on why B2B landing pages fail to convert.
The combination of tighter keyword control, better bidding signals, and dedicated landing pages dropped the cold search CPL from $312 to $168 within 60 days. The retargeting CPL came in at $74, which was expected given the audience quality. Blended across all paid channels, the CPL settled at $143 by day 90.
Results and What Drove Them
By the end of the 90-day period, the client's total paid media spend had actually decreased by 11%, from $28,000 to $24,900 per month, because we eliminated the wasted allocation to PMax and low-intent keywords. SQLs increased from 31 per month to 43 per month, a 38% lift. Pipeline value generated from paid channels increased by 61%, from an average of $190,000 per month to $307,000, because the higher-quality leads from LinkedIn-influenced paths had larger deal sizes.
The single most impactful change was the offline conversion import. Before it, Smart Bidding was optimising for form fills from anyone, including competitors, students, and early-stage researchers with no budget. After it, the algorithm had a clear signal: optimise for the specific query and audience patterns that produce six-figure deals. That one technical fix accounts for roughly half of the CPL improvement on its own.
The lesson here is not that LinkedIn is always undervalued or that Google Ads is always over-credited. The lesson is that any account running last-click attribution across multiple channels is making budget decisions based on incomplete data. Fixing the measurement layer first, before touching bids or budgets, is the sequencing that makes everything else work. If you are seeing similar patterns in your own numbers, the diagnostic process we use is covered in detail in our guide to why Google Ads stops generating quality leads.