How should B2B marketing be measured in the AI age?

Why legacy metrics break

B2B marketing measurement was built for a world where the buyer's journey started with a Google search, proceeded through a content download, and ended with a form fill. Every step was instrumented. Sessions, bounce rate, MQL count, SQL conversion, pipeline contribution, all of it worked because the buyer was on your property before the first meaningful conversation.

That world is eroding. Roughly 40% of B2B category research in 2026 now starts inside ChatGPT, Claude, Gemini, or Perplexity, and the buyer often arrives at your site already with a shortlist, having read an LLM-generated answer that cited you (or did not). The buyer's most valuable research step is invisible to your analytics stack. Legacy KPIs are not wrong; they are measuring a smaller and smaller slice of the actual buyer journey.

The five metrics that replace the old ones

1. Citation share

The percentage of LLM responses to your tracked category questions that cite your entity by name. Measured monthly across ChatGPT, Claude, Gemini, and Perplexity on 3 to 10 tracked questions. Tracked alongside citation position (first-mentioned vs later) and co-citation set (which other entities appear with you). Citation share is the closest 2026 analog to top-of-funnel impressions, but it is a much higher-quality signal because it reflects what a buyer actually reads, not what a model served on an ad network.

2. ICP follower composition

Of the net new followers you added on the founder's LinkedIn (and X, and other social surfaces) in the last month, what percentage hold titles that map to your Ideal Customer Profile? For most B2B companies, this is the single most predictive metric of future pipeline, more predictive than total follower count by a factor of 3 to 5. If ICP share is under 30%, the founder's POV is not reaching the right audience even if the engagement numbers look healthy.

3. Inbound first-call volume

The number of first calls booked in the last month that originated from inbound replies, mentions, comments, or DMs on founder-led or company-led surfaces. This replaces MQL count as the primary pipeline quality metric because it measures buyer-initiated interest, which converts at 3 to 8x the rate of outbound-initiated interest across most B2B categories. Track source (which post, which podcast, which publication) so you can connect pipeline back to POV.

4. Sourced pipeline attribution

Pipeline sourced from inbound and founder-led channels, tracked through to closed revenue on the same multi-quarter cadence as direct sales. This is the finance-facing version of 'is marketing working?'. For companies running founder-led marketing with AEO layered on top, sourced pipeline as a share of total pipeline typically moves from 10 to 20% to 40 to 60% over 12 to 18 months.

5. Sales cycle time on inbound vs outbound

A less-watched but highly-diagnostic metric. In most B2B categories, inbound deals that originated from the founder's published POV close 30 to 45% faster than outbound deals of similar size. This is because the buyer arrives having already processed your POV, meaning discovery is shorter, objections are already addressed, and internal selling is easier. Tracking cycle-time delta between inbound and outbound gives the marketing function a clean KPI that finance understands.

What to keep from the legacy dashboard

Not everything should be retired. Three legacy metrics still matter:

• Brand search volume. When a buyer searches your company name on Google or types it into ChatGPT, they are moving down the funnel. Brand query volume is still a credible mid-funnel signal.
• Pipeline velocity from inbound leads. Tracking how quickly inbound leads progress to first call, then to qualified opportunity, remains the cleanest single signal of marketing-to-sales handoff quality.
• Customer LTV and net revenue retention. Marketing that brings in the wrong customers can be invisible in pipeline metrics but visible in expansion and churn. NRR is the ultimate backstop.

What to retire

• MQL count as a primary KPI. MQL thresholds were a bandaid for the inability to measure buyer intent. Inbound first-call volume is the higher-quality replacement.
• Session count and time-on-site as primary content metrics. LLMs cite content that buyers never visit. Traffic under-counts real influence.
• Last-click attribution. By 2026, most meaningful marketing influence happens before the attributed click. Run multi-touch if you must attribute, but bias toward first-touch for funnel education.
• CPM-denominated paid campaign reporting. Paid can still play a role, but it should be measured on sourced pipeline and cycle-time delta, not CPM or CTR.

Instrumenting for the new dashboard

Four pieces of infrastructure do most of the work:

1. 01A citation tracker that queries ChatGPT, Claude, Gemini, and Perplexity monthly on your tracked questions and records entity mentions, position, and co-citations. Several tools in the market do this; Revintl runs a custom one for engagements. The key is consistent month-over-month methodology.
2. 02A follower composition classifier that reads new LinkedIn followers on the founder's account, infers title and company, and buckets them into ICP vs non-ICP. Can be run weekly in a spreadsheet; scales to a small internal tool at volume.
3. 03An inbound intake pipeline that captures reply-originated first calls separately from outbound-originated ones in your CRM. A simple origination field in Salesforce or HubSpot is enough to start.
4. 04A content-to-pipeline link that records which posts, podcasts, or publications an inbound reply cited. This is the data that lets you connect POV to revenue, and it is almost always missing in B2B CRMs. A freeform note field captured during the first call is fine.

How to introduce the new metrics without losing finance's trust

Finance will reasonably ask why you are changing the scoreboard. Three moves make the transition credible:

• Run the new dashboard in parallel with the legacy one for two full quarters. Show both sets of metrics side by side, explain the drift between them, and let finance see the story before you retire anything.
• Tie the new metrics to sourced pipeline and closed revenue, not to themselves. A dashboard that reports citation share in isolation is vulnerable. A dashboard that reports citation share alongside its contribution to inbound first calls, sourced pipeline, and closed revenue tells a connected story.
• Be honest about lag. Citation share moves on 3 to 9 month lags. Sourced pipeline attribution takes another quarter to mature. Set the expectation that the new dashboard will look quiet for the first two quarters and pay off in quarters three and four.

A reporting template for 2026

Most working B2B marketing dashboards in 2026 are structured in four tiers:

1. 01Outcome metrics (executive view): sourced pipeline, cycle-time delta on inbound vs outbound, pipeline-to-plan ratio.
2. 02Intent metrics (team view): inbound first-call volume, inbound origination by channel (founder LinkedIn, podcast, publication, community), POV-to-pipeline tracing.
3. 03Authority metrics (compound view): citation share on tracked questions across the four major LLMs, ICP follower composition, share of category voice on tracked keywords.
4. 04Output metrics (operational view): post cadence, podcast appearances shipped, publication placements, engagement per post, response SLA on inbound replies.

Reviewed weekly for output and intent. Monthly for authority. Quarterly for outcomes.

The short version

Legacy B2B marketing metrics were built for a Google-first world. In 2026, 40% of category research happens inside an LLM before the buyer reaches your site, which makes your analytics invisible to the most important step. The fix is to measure entity-level authority (citation share), audience quality (ICP follower composition), and outcome (inbound first-call volume and sourced pipeline), and to run the new dashboard in parallel with the legacy one until finance and the team are confident in the story.