Form fills look fine, pipeline doesn't move. The six-step fix sequence, in order: offline conversions, value-based bidding, audience exclusions, form filtering, match types, lead form match-quality.
B2B form fills look fine but pipeline does not move because Smart Bidding optimizes for form_complete, not closed revenue. The fix sequence matters. Do these in order:
Expect 30-50 percent fewer total form fills and a 2-3x higher pipeline conversion rate within 30-60 days. The drop in raw fills is the point, not a regression. As Cometly's 2026 fix-playbook notes, a campaign with 50 leads at a 40 percent close rate beats one with 100 leads at 5 percent on every revenue line (Cometly, 2026).
Three systems track different scoreboards on the same campaign, and each one looks fine in isolation. The gaps between them produce the leads sales rejects.
Reason 1: Smart Bidding sees one signal, sales tracks another. Google Ads counts every form_complete as one conversion of equal weight. Sales tracks MQL, SQL, opportunity, closed-won. When 65 percent of B2B buyers start their journey on Google search (The Marketing Blender, 2026), the volume of form_complete events the algorithm sees is enough to feel successful. The downstream stages tell a different story.
Reason 2: B2B intent is harder to read from a single query. A search for "CRM software" can mean a senior buyer evaluating tools (high value), a junior researcher building a longlist (medium), a student writing a case study (zero), or a freelancer wanting a free trial (zero). Same query, four very different leads. Without feedback, Smart Bidding bids the same on all four.
Reason 3: The form does not qualify. A four-field form (name, email, company, phone) accepts anyone. No business-email validation, no company-size gate, no qualifying questions. Gmail addresses and freelance consultants flow through unfiltered, then sit in sales queues as rejected.
If you have not isolated the root cause yet, start with our breakdown of why Performance Max fails in B2B marketing. It walks through the diagnose side of the same problem. This article is the fix side.
Without offline conversion imports, Google has no way to know which form fills became revenue. Configure this before changing anything else. It is the single highest-leverage fix in the playbook, and it assumes your baseline Google Ads conversion tracking is already firing correctly. If it is not, fix the basics first.
Google now calls the modern version "Enhanced conversions for leads" (Google Ads support, 2026). The legacy term "offline conversion imports" still works in most accounts. Either way, the mechanic is the same:
A reasonable per-stage value schema (illustrative, adjust to your average contract value):
The schema does not have to match true economics perfectly. It has to be directionally accurate and consistent. Smart Bidding learns the gradient between stages, not the absolute dollar amounts.
Implementation paths by stack: HubSpot has a native Google Ads integration that handles GCLID and stage uploads in a few clicks. Salesforce uses the Marketing Cloud connector or a third-party (Zapier, LeadsBridge). Engineering teams can use the Google Ads API or the modern Conversions API directly.
Only about 13 percent of businesses send any offline conversion feedback back to Google Ads (Pete Bowen, 2025). That gap is most of why so many B2B accounts feel Google Ads does not work for them. Closing it typically reduces average cost-per-lead by around 31 percent (The Marketing Blender, 2026).
Once offline conversions are flowing, switch the bid strategy. Maximize Conversions optimizes for count. Maximize Conversion Value optimizes for revenue. That single change reshapes what Smart Bidding pursues.
Migration sequence that works: run Maximize Conversions for 30 days to build a baseline. Switch to Maximize Conversion Value. Add Target ROAS only after another 30 days of value data. The full ramp takes 60-90 days, in line with Cometly's 2-4 week algorithm learning windows applied across two strategy phases (Cometly, 2026).
Counter-example we see often: B2B accounts flip directly to Target ROAS with an aspirational target. The system cannot find conversions at that ROAS, spend collapses, the team blames the algorithm. Run it loose first.
B2B campaigns leak budget to job-seekers, freelancers, students, and competitors auditing your offer. Exclude them at campaign level or, better, at account level via a shared exclusion list.
| Audience exclusion | Source signal in Google Ads | Why exclude in B2B | Typical spend reclaim |
|---|---|---|---|
| Job-seekers | In-market: Employment | Search company name plus careers, hit retargeting | 5-15% |
| Freelancers / self-employed | Affinity: Small Office Workers (selective) | Want free trials, no buying authority | 3-8% |
| Students | Detailed demographics: Education Status | Researchers, case-study writers, zero pipeline value | 2-5% |
| Competitors / vendors | IP exclusions plus competitor domain lists | Audit your offer, never buy | 1-3% |
| Existing customers (paid acquisition) | Customer Match list as exclusion | Renewal flow lives elsewhere | 5-12% |
| In-house employees | Customer Match list (employee emails) | Test traffic distorts the signal | 1-2% |
Implementation tactic: stack exclusions at the account level (Tools, Shared library, Audience manager) so they apply to every campaign. Per-campaign exclusion lists drift out of sync quickly. Account-level lists do not.
One caveat. Customer Match list size has to hit Google's match threshold (typically 1,000+ users) before it activates as an exclusion. Smaller B2B accounts may not have list density to use this lever yet. Build the list anyway, it will activate once volume catches up.
The form is your cheapest qualifier. Every field you add filters before sales touches the lead. The marginal cost of a longer form is a 15-30 percent drop in form_complete count. The marginal benefit is a 2-3x lift in sales-accepted rate. The math almost always favors longer forms in B2B.
The five form changes that produce the largest filtering effect:
The counter-balance honesty: longer forms cut form_complete count. That is the point. You are not optimizing for forms, you are optimizing for sales-accepted leads. After Fix #1 is live, Smart Bidding accepts the lower form count and bids harder on qualified leads.
Broad match in B2B without aggressive negative keyword maintenance is a budget leak. Most accounts cannot sustain the daily review cadence broad match requires, so the practical default is tighter.
Match-type rule of thumb for B2B:
The weekly negative-keyword routine that catches 80 percent of leakage in 30 minutes:
For the full procedural depth on match-type strategy, our breakdown of Google Ads keyword match types walks through the trade-offs with examples.
Google now reports a match-quality score for lead form extensions. The score is the platform telling you which ad groups are attracting wrong-intent leads. Most B2B accounts ignore the signal, which is a missed opportunity since Google is essentially flagging waste for you.
Where to find the score: Google Ads, Campaigns, select campaign, Ads & assets, Lead forms, expand the row for the Match quality column. The feature went into general availability in 2025.
Threshold rules of thumb:
Pair this signal with CRM-side MQL conversion rate per ad group. The two metrics correlate. Low match-quality typically precedes low MQL rate by 14-21 days, which means you get a leading indicator instead of a lagging one.
That CAC delta sits inside Cometly's reported 20-40 percent range for closed-loop B2B teams (Cometly, 2026). It is unusual only in how cleanly it shows up when the order of operations is right.
Sometimes the campaign architecture is the actual problem. Patching offline conversions on top of broken structure wastes 6-8 weeks and produces a frustrated team. Three signals say rebuild instead of patch:
Signal 1: Campaign-type sprawl. Performance Max plus Search plus Display plus Demand Gen all running broad with overlapping audiences. Audience signals leak across campaigns. The fix is isolation by intent: Search for commercial, PMax only after a Search baseline exists for B2B. The diagnose side of this is in our Performance Max in B2B breakdown.
Signal 2: Ad copy and landing page promise different offers. A perfect offline conversion setup cannot fix copy-product mismatch. Intent-matched landing pages produce conversion rates up to 4x higher than generic homepages (The Marketing Blender, 2026), so the landing page is part of the qualification chain, not a separate problem.
Signal 3: CRM data hygiene is too weak for OCI to work. If sales does not update stages reliably or contacts are duplicated and orphaned, the signal you send Google is noise. Pause the OCI rollout, fix the CRM operationally for 30-60 days, then re-attempt. As Ramsey Sanchez puts it in The Marketing Blender's 2026 B2B roundup, do not just focus on the cost per click, you need to be focusing on optimizing for lead and lead quality. Quality of the data feeding the loop matters more than quantity.
If two or more of these signals are present, sequence the rebuild before the OCI install. If only one is present, fix it in parallel.
How long until offline conversion imports actually change Smart Bidding behavior? 14 days minimum after stable uploads begin. Google needs that window for the algorithm to ingest stage data and recalibrate. Some accounts see meaningful shifts at 21-30 days, in line with Cometly's reported 2-4 week algorithm learning windows (Cometly, 2026).
What if my sales cycle is 9 months? Will OCI still work? Yes, with one adjustment: use pipeline-stage values (MQL, SQL, Opportunity) rather than waiting for closed-won. Smart Bidding learns from the early stages and uses them as a leading indicator. Final closed-won uploads refine the model over quarters but do not have to be the input signal.
Should we use Google Ads lead form extensions or send traffic to a landing page form? Lead form extensions yield more leads at lower cost per lead, but lower average quality. Landing page forms allow custom qualification fields. For B2B, landing page forms with qualifying questions almost always win on pipeline-valued CAC, even at lower volume.
How do we handle privacy and consent with offline conversion imports? Capture explicit consent at form submission for data sharing with Google. For EU traffic, run consent mode v2. Without proper consent, OCI uploads should not transmit user-level identifiers, and Google may silently drop them.
Can negative keywords break Smart Bidding's learning? No. Adding negatives does not reset the learning phase. Changing bid strategy or shifting budget more than 20 percent does. Negative keyword maintenance is safe to do weekly without disrupting the algorithm.
Even with all six fixes in place, lead quality drifts. CRM API tokens expire. Sales reps stop updating stages. A new audience signal goes live and skews mid-funnel. Aegis (the B6 risk-review agent) runs weekly drift detection on MQL conversion rate by ad group, lead form match-quality trend, and OCI upload health. The detection fires before budget burns rather than after.
Once direction is confirmed by a human, Maximus orchestrates the apply: pause underperforming ad groups, push negative keyword updates, refresh OCI configurations. The human stays in approval, the agents do the click work. The same teams who land in Cometly's 20-40 percent CAC reduction band (Cometly, 2026) are the ones who automate the maintenance loop, not just the initial fix.
See how Aegis and Maximus run weekly drift detection and apply across your Google Ads accounts.
See Kampaio Pricing