Most B2B marketing teams do not have a targeting problem or a creative problem. They have an infrastructure problem - and they cannot see it, because the dashboards keep loading.
Before you rewrite ads, rebuild the plumbing.
The Premise
Performance marketing in B2B is not a single discipline. It is four layers of infrastructure stacked on top of each other: how you tag traffic, how you attribute revenue, how leads are routed to sales, and how the outcome flows back into the system that spends the money. When any layer is weak, everything above it inherits the noise.
Teams usually notice the top of the stack first - the creative, the targeting, the budget split - because that is what is visible in the platform UI. But when a campaign "underperforms," the cause is almost never the campaign. It is a UTM that never landed in the CRM, an offline conversion that was never uploaded, a lead that sat unrouted for nine days, or a "closed lost" reason field that nobody fills in.
You do not have a marketing performance problem. You have a marketing infrastructure problem the dashboards are too polite to name.
The Four Infrastructure Layers
The audit below covers the four layers in order of causality. Fixing layer 1 without layer 2 gives you clean data no one uses. Fixing layer 3 without layer 1 routes garbage faster. Work top-down.
- Tracking & Tagging - the ledger. If it is wrong here, nothing downstream is trustworthy.
- Attribution & Reporting - how spend is connected to revenue, including offline conversions and consent-gated signal.
- Routing & Handoff - the operational moment where marketing hands a person to sales. Most pipeline dies here.
- Feedback & Learning - whether the outcome ever reaches the channel that spent the money. Without it, optimisation is fiction.
1. Tracking & Tagging
The tracking layer is the ledger every other decision is built on. Two questions matter: can you tell where a session came from, and does that answer still hold when the record reaches your CRM?
Tracking checklist (10 points)
- A written UTM convention exists and every paid channel follows it - lowercase, no spaces, controlled vocabularies for source/medium/campaign.
- utm_content and utm_term are used consistently for ad-level and keyword-level granularity, not left to platform auto-tagging alone.
- GA4 is deployed via GTM, with server-side tagging or Enhanced Measurement configured deliberately, not by default.
- Consent Mode v2 is live with all four signals (ad_storage, ad_user_data, ad_personalization, analytics_storage) and default state matches your legal posture.
- First-party identifiers (email hash, user_id) are captured on form submit and passed to GA4, Google Ads and LinkedIn where consent allows.
- LinkedIn Insight Tag, Meta Pixel and Google Ads tag fire only after marketing consent is granted, not on page load.
- UTM parameters are written to hidden fields on every form and persisted to the CRM contact record - not dropped at submission.
- A cross-domain measurement plan exists if traffic moves between marketing site, product, and a separate booking or checkout host.
- 404s, redirects and campaign landing pages are monitored - a broken UTM link is a silent budget leak.
- You can answer 'where did this closed-won deal originate?' from the CRM alone, without opening an ad platform.
2. Attribution & Reporting
Attribution in B2B is not last-click and not a model - it is a decision about which signals you trust the ad platforms to optimise against. Since ATT, iOS 17 link stripping, and Consent Mode, the reliable path is fewer, higher-quality conversions fed back to the platforms deliberately.
Attribution checklist (10 points)
- A single source of truth exists for pipeline and revenue - CRM, not a spreadsheet, not a BI tool disconnected from the CRM.
- Offline conversions (MQL, SQL, Opportunity, Closed Won) are uploaded to Google Ads and LinkedIn on a defined cadence - daily or better.
- Enhanced Conversions for Leads is configured in Google Ads with hashed email/phone from the CRM, not just from the form.
- LinkedIn Conversions API is live where the account qualifies, replacing pixel-only signal.
- Attribution windows are documented per channel and match the actual sales cycle (30/60/90 day is not one-size-fits-all).
- Self-reported attribution ('How did you hear about us?') is captured on the demo form and reconciled quarterly with platform-reported data.
- A weekly report shows spend, MQLs, SQLs, Opps and Closed Won by channel - not just CPC and CTR.
- The reporting distinguishes new-business pipeline from expansion/renewal, and paid from organic influence.
- Someone owns the reconciliation between CRM totals and BI totals - and the delta is under 5%.
- The team knows the CAC and payback period per channel, and can defend both numbers with primary data.
3. Routing & Handoff
Routing is the physical moment where marketing hands a person to sales. It is boring, under-instrumented, and where most pipeline dies. Speed to lead, ownership, and lifecycle discipline decide whether a working funnel actually produces meetings.
Routing checklist (10 points)
- Lifecycle stages are defined in writing with entry and exit criteria - not left to sales reps' interpretation.
- An MQL-to-SQL SLA exists and is measured: median first-touch time, percentage responded within 5 minutes, percentage abandoned.
- Round-robin, territory or account-based routing rules are automated in the CRM, not manual assignment via Slack.
- Every inbound demo request triggers an immediate notification to the assigned rep - email, Slack and mobile push, not just an email.
- Duplicate detection runs on email and company domain before a new lead is created - no split pipeline across two records.
- Lead scoring exists and is calibrated against actual SQL and Closed Won outcomes, not vibes - and it is reviewed quarterly.
- 'Working' vs 'not working' status is visible in the CRM within 48 hours of assignment; stalled leads route to a nurture path automatically.
- Disqualification reasons are a required, controlled-vocabulary field on lead status change - not a free text.
- Contact data enrichment (firmographic + role) runs on lead creation, not manually by the SDR.
- A daily or weekly funnel report shows conversion rates between every stage - and someone is accountable for each rate.
4. Feedback & Learning
The last layer is the one that closes the loop: whether the outcome of a lead ever reaches the platform that spent the money to acquire it. Without this, ad platforms optimise for form fills - and you get more form fills instead of more revenue.
Feedback checklist (10 points)
- Google Ads and LinkedIn campaigns bid on SQL or Opportunity events, not on form submits, wherever conversion volume supports it.
- A negative-signal pipeline exists: disqualified and closed-lost leads are excluded or suppressed from lookalikes and retargeting.
- Closed-won accounts are excluded from prospecting campaigns automatically, via matched audiences or company-list exclusions.
- Sales notes and disqualification reasons are read by the marketing team on a defined cadence (weekly at minimum).
- Marketing sits in on a sample of sales calls or has access to call recordings monthly - qualitative signal beats another dashboard.
- Content and creative are informed by 'closed lost' reason distribution, not only by 'closed won' testimonials.
- A/B tests run on landing pages against downstream metrics (SQL, pipeline) - not only against form-fill rate.
- Budget reallocation between channels happens on a defined cadence (monthly or better) using the same reconciled numbers everyone else uses.
- New campaign proposals include a hypothesis, a target funnel-stage metric, and a kill criterion - not just a creative brief.
- The team can name three things it stopped doing in the last quarter because the data said so. If it cannot, the feedback layer is broken.
How to Score the Audit
Count the boxes you can check honestly. The distribution tells you where to start - not the total.
- 35-40: mature infrastructure. Optimise creative, targeting and offer. Marginal gains are real.
- 25-34: functional but leaky. One or two layers are dragging the rest down; identify which and fix in sequence.
- 15-24: partial infrastructure. Any spend increase here amplifies noise, not results. Pause new experiments and rebuild.
- Below 15: the dashboards are decorative. Stop optimising and start instrumenting - in that order.
The Fix Sequence
The order matters. Every fix upstream compounds; every fix downstream is temporary until the upstream fix lands.
- Tracking first. Nothing else is trustworthy without it. A week of UTM governance work saves a quarter of misattributed budget.
- Routing second. A fast, disciplined handoff turns existing MQL volume into more SQLs without any new spend.
- Feedback third. Once tracking and routing hold, feeding real outcomes back to the ad platforms is the single largest lever on CAC.
- Attribution reporting last. A report built on the first three layers is worth reading. Built on none of them, it is theatre.
Fix the plumbing before you paint the walls. Every layer above depends on every layer below.
A working infrastructure is not glamorous. It does not photograph well for a board slide. But it is the difference between marketing that is active and marketing that produces revenue - and in specialist B2B markets, that difference is the entire game.
Further reading
Google Ads Enhanced Conversions for Leads documentation; LinkedIn Conversions API reference; Google Consent Mode v2 (mandatory March 2024); Forrester 2024 Sales & Marketing Alignment Survey; HubSpot lifecycle stage and lead scoring documentation; NM Insight, "Your CRM Is the Real Marketing Tool" (Issue 03).
NM Editorial Team
Analysis and market intelligence from NM Insight - a Berlin-based B2B performance marketing consultancy.