A B2B payments infrastructure company came to us in Q1 2026 with a familiar problem: their Google and LinkedIn campaigns were generating roughly 180 leads per month, but sales was closing fewer than 4% of them as SQLs. The pipeline looked healthy on a dashboard and hollow in reality. Over 11 weeks, we cut their junk lead rate by 68% and doubled their SQL rate from 4% to 9.1%, without reducing total monthly spend by a single dollar.
The Diagnosis: Volume Was Hiding a Targeting Collapse
The client's Google Ads account was running broad match keywords like 'payment processing software' and 'corporate payments platform' without a meaningful negative keyword list. Those terms were pulling in SMB searchers, freelancers, and even job seekers at a clip that accounted for roughly 55% of form submissions. LinkedIn was no cleaner: their Sponsored Content campaigns were targeting by job function and seniority, but the company size filter was set to 'all', which let in sole traders and two-person startups that would never qualify under their minimum contract threshold of $2,000/month.
On the form side, the lead capture asked only for a name, email, and company name. There was no company size selector, no use-case qualifier, and no monthly volume field. Sales was spending an average of 22 minutes per lead on discovery calls before confirming disqualification. That is a significant hidden cost that never appears in a CPL report. Understanding how form structure drives pipeline quality is something we cover in detail in our analysis of why B2B lead forms get clicks but not leads.
Week 1-3: Audience Surgery on LinkedIn
The first intervention was purely structural on LinkedIn. We tightened company size to 51-5,000 employees, added seniority filters for Director, VP, and C-Suite, and layered in job titles specific to treasury, finance operations, and payment infrastructure. We also excluded industries with near-zero historical close rates: staffing, education, and non-profit. The audience shrank by about 40%, but the relevance score on the Sponsored Content units rose from 6 to 9 within the first two weeks as engagement rates improved.
We simultaneously paused the two lowest-performing ad sets, which together consumed 28% of the monthly LinkedIn budget, and reallocated that spend to a Thought Leader Ads format featuring the client's CFO discussing reconciliation pain points. CTR on that format came in at 0.71%, compared to 0.29% for the static image ads it replaced. The important nuance here is that the thought leader format also self-selects: prospects who engage with a CFO-narrated pain point video are demonstrably more senior than those clicking a generic 'book a demo' banner.
Week 2-4: Negative Keywords and Match Type Restructure on Google
On the Google side, we ran a 30-day search term report and found 214 unique queries that had generated at least one click but zero pipeline. We built a negative keyword list of 180 terms across three themes: SMB-intent modifiers ('small business', 'freelance', 'cheap', 'free'), informational queries ('what is', 'how does', 'tutorial'), and job-seeker terms ('careers', 'salary', 'review on glassdoor'). This alone reduced irrelevant impressions by 34% in the first two weeks after implementation.
We also moved the three highest-spend ad groups from broad match to phrase and exact match, with a small budget ring-fenced for a controlled broad match test with audience-layered targeting applied. Google's own documentation on match types makes clear that broad match works best when paired with smart bidding and strong audience signals, but this account had neither in place before we arrived. The restructure cut CPC by 18% on exact match terms while improving average position for the high-intent queries that actually mattered.
Week 4-7: Form Friction as a Qualification Tool
The form redesign is where the SQL rate moved most sharply. We added three qualifying fields: monthly transaction volume (with a dropdown starting at 'under $10K' through 'over $500K'), company headcount (under 50 / 50-500 / 500+), and a single-question use case selector covering 'cross-border payments', 'accounts payable automation', and 'reconciliation'. We also changed the CTA copy from 'Get a Demo' to 'Check If You Qualify', which set an expectation of selectivity rather than open access.
Form completion rate dropped from 6.2% to 4.9% after the change. That is a 21% reduction in raw lead volume, and it was entirely intentional. The leads that did complete the longer form converted to SQL at 11.4% in weeks 6 and 7, compared to 4% before. Sales reported that discovery calls were shorter and more productive because reps arrived with real context. This approach reflects a broader principle: using intent signals to qualify B2B leads before they ever reach your CRM saves more revenue than chasing volume.
Week 7-11: Closed-Loop Reporting and Budget Reallocation
With the CRM now receiving structured qualification data from the new form fields, we built a simple lead scoring layer inside HubSpot: companies over 200 employees with transaction volumes above $50K/month scored 80+, triggering an immediate sales notification rather than a 48-hour nurture sequence. That alone cut the average time-to-first-sales-contact from 51 hours to under 4 hours for high-score leads. HubSpot's lead scoring model made the routing logic straightforward to implement without custom development.
By week 11, the numbers had settled as follows: junk leads (defined as disqualified at the first sales touchpoint) fell from 72% of all submissions to 23%. SQL rate moved from 4% to 9.1%. Revenue pipeline attributed to the two channels increased by $340K in the 60-day period following full implementation, on an unchanged monthly ad spend of $18,500. The client's cost-per-SQL dropped from approximately $1,620 to $510. If your own campaigns are generating similar disconnects between lead volume and pipeline output, a structured growth audit is often the fastest way to identify exactly where the breakage is occurring before committing to further spend increases.
What This Case Teaches About Lead Quality at Scale
The core lesson is that lead quality problems almost never have a single cause. In this case, three separate layers were each contributing independently: audience targeting that was too broad, keyword match types that pulled in the wrong intent, and a frictionless form that offered no natural filter. Fixing only one layer would have produced a partial improvement. Fixing all three, in the right sequence, produced a compounding effect that showed up clearly in the SQL rate within two months.
The sequence also matters. Audience and keyword fixes come first because they change who sees your ads. Form changes come second because they filter who actually submits. Closed-loop reporting comes third because it tells you whether your definition of a qualified lead is actually predictive of revenue. Skipping to form optimization without fixing the traffic source is a common mistake: you end up filtering a pool that is already contaminated at the source, which means you lose volume without recovering quality.
B2B fintech is a harder environment than most for lead quality because the keywords are expensive (the client's average CPCs on competitive terms ran $18-$34), the buying committee is large, and the sales cycle runs 60-120 days. That combination makes every junk lead disproportionately costly. The mechanics described here, negative keyword depth, audience size filters, form-based qualification, and CRM routing, apply across most B2B verticals, but the urgency of getting them right is highest where CPC and sales cycle length both run long.