A company selling fleet telematics to logistics operators held its monthly performance review. The media manager opened with 14 million impressions and an engagement rate up a third on the previous month. The sales director answered with pipeline, which was flat. The finance analyst quoted cost per lead, up 20 per cent on the quarter. The meeting ran forty minutes without reaching a decision. Senior management outside the media team left with the impression that marketing could not explain its own spend. All three numbers were accurate. Each number came from a different layer of measurement: the media manager quoted Delivery and Engagement metrics, the finance analyst quoted a Lead metric, and the sales director quoted a Pipeline metric.

Every paid media metric sits at one of five layers between spend and revenue: Delivery, Engagement, Lead, Pipeline and Revenue. The meeting above stalled because each person was answering a different question. This post covers what each layer proves and cannot prove, which layers to use for each kind of channel, why the Lead layer causes most reporting mistakes, why the Pipeline and Revenue layers exist only in the CRM, and how to build a one-page report senior management can trust. The series hub, How B2B Media Actually Works (and What It Cannot Do), covers the research behind the plan as a whole: the 95:5 rule, the buying committee, and demand creation versus demand capture.

The Five Layers of Paid Media Metrics

Every metric a media plan can report belongs to one of five layers, running from evidence the ads ran through to closed revenue. When two people disagree about media performance, write each quoted metric's layer beside it: the disagreement is usually about which layer matters, not about either number being wrong.

The Delivery layer

Delivery metrics are impressions (one loading of an ad on a page or in a feed), reach (the number of people or accounts that received at least one impression), frequency (the average number of impressions per person or account reached) and viewability (the share of impressions that appeared on screen long enough to be seen). The Delivery layer proves the media ran and who the media ran against. The Delivery layer cannot prove anyone noticed.

The Engagement layer

Engagement metrics are clicks, view time, video completions (a video played through to the end) and engagement rate (interactions divided by impressions). The Engagement layer proves someone noticed the ad and responded to it. The Engagement layer cannot prove the person who noticed has any intention of buying.

The Lead layer

A lead is a person who has identified themselves, almost always by completing a form. Lead metrics are form completions, contacts added and cost per lead (spend divided by the number of leads). The Lead layer proves someone was willing to hand over a name and an email address in exchange for something. The Lead layer cannot prove the person is a buyer, because people complete forms for many reasons and most of those reasons are not a purchase.

The Pipeline layer

Pipeline metrics are qualified opportunities (potential deals sales has reviewed and accepted as worth pursuing), cost per qualified opportunity, and pipeline value (the combined potential value of the open opportunities a channel contributed to). The Pipeline layer proves the CRM, the customer relationship management system where sales records accounts, contacts and deals, connects the spend to potential revenue. The Pipeline layer cannot prove the potential revenue will close.

The Revenue layer

Revenue metrics are closed business, win rate (the share of opportunities that end in a sale) and revenue per channel. The Revenue layer proves deals closed and revenue arrived. The Revenue layer cannot prove the media caused the revenue, because a closed deal was shaped by sales work, product, pricing and timing alongside every ad the account ever saw.

In my experience, the fastest fix for a stalled reporting meeting is writing the layer number beside every metric on the page. Half the disagreements end there, because the people arguing discover they were answering different questions.

The layers connect weakly. Strong delivery does not guarantee engagement, and engagement does not guarantee leads. Lead volume says surprisingly little about pipeline, and pipeline still has to survive sales execution before pipeline becomes revenue. Each layer is a leading indicator of the next, with weakening correlation at every step, which is why a report needs to show more than one layer.

Match the Metric to the Channel's Job

Each channel should be judged at the layers that match the channel's job. Demand creation channels build memory among buyers who are not yet looking for a solution. Demand capture channels intercept buyers who are already looking. Creation pays out over quarters and capture pays out within the quarter, so the two kinds of channel must be judged at different layers.

Demand creation channels, chiefly the channels covered in Paid Social in B2B: Reaching Buyers Who Are Not Looking and Video Advertising in B2B: The Strongest Creation Format, on a Lag, should be judged at the Delivery and Engagement layers in the quarter of the spend, and at the Pipeline and Revenue layers two or more quarters later. The reason is the lag. Les Binet and Peter Field's analysis of the IPA effectiveness databank in The Long and the Short of It shows that brand-building activity pays out over long periods, while activation activity pays out quickly and fades. A creation campaign doing its job well in March shows up as delivery and engagement in March and as pipeline months later. Judging the campaign on leads, pipeline or revenue in March measures results the campaign has not had time to produce.

Demand capture channels, chiefly paid search and the review sites, can be judged at the Lead, Pipeline and Revenue layers in the quarter of the spend. The buyers those channels reach are already in motion, so the leads convert or go cold within weeks, and the CRM can connect the spend to pipeline in the same reporting period.

Judging a demand creation channel at the Lead layer in the quarter of the spend is the single most common reporting error in B2B. The error is easy to commit because cost per lead sits at the Lead layer, and cost per lead is the number most monthly reports are built around. The Ehrenberg-Bass Institute's research in How B2B Brands Grow holds that around 95 per cent of category buyers are out of market in any given period. A social or video campaign built to reach those out-of-market buyers will produce few leads this quarter by design, and a Lead-layer verdict records the design as failure. The verdict then moves budget away from the channels building next year's pipeline and toward the channels harvesting demand somebody else created.

The Lead Layer Is Where Reporting Goes Wrong

The Lead layer deserves its own section, because the Lead layer is where most B2B media reporting fails. Three properties of the layer cause the failure.

The first property: cost per lead compares different events as if the events were equal. A demo request is a hand raised by a buyer asking to see the product. A syndication download is a contact record generated because a publisher's subscriber accepted a document, often without remembering which company sponsored the document, a mechanism examined in Content Syndication in B2B: Buying Names, Not Buyers. Both events arrive in the CRM as one lead each, and cost per lead prices both at face value. Ranking channels on cost per lead therefore compares a buying signal against a purchased name, and the purchased name ranks higher because the purchased name costs less.

The second property: the Lead layer is the easiest layer to inflate. Any channel can cut its cost per lead by lowering the commitment the form asks for. Put a generic document behind a two-field form and leads multiply while the value of each lead collapses. The Delivery layer is hard to fake at scale, and the Pipeline layer is policed by salespeople who accept or reject each opportunity. The Lead layer sits between the two with no equivalent check, so cheap forms flood the CRM wherever cost per lead is the number being managed.

The third property: the Lead layer is popular because lead counts arrive fast and look precise in a monthly report. A lead count moves within days of a budget change, comes with names attached, and fits neatly on a chart, while pipeline takes quarters to form. The speed makes the lead count useful for operating a campaign week to week. The speed does nothing to make the lead count a fair verdict on a channel.

The table below is a worked example for a company selling contract management software to mid-market legal teams. The company runs four lead sources, ranks the sources by cost per lead, then re-ranks the same sources by cost per qualified opportunity once each source's leads have had two quarters to mature.

Example figures for illustration. The ranking reverses completely: the source with the cheapest leads produces the most expensive qualified opportunities, and the source with the most expensive leads produces the cheapest. A team managed on cost per lead would move budget toward the top row and starve the bottom row.

The reversal happens because each lead type carries a different level of intent, and cost per lead is blind to intent. Cost per qualified opportunity, read after enough lag for leads to mature, is the comparison metric that survives scrutiny. One caveat applies. The paid social row prices only that channel's lead forms. The channel's main contribution is memory among buyers who are not yet in market, and no lead-layer table can price that contribution.

Crossing the Layers Needs the CRM

Platform reporting stops at the Engagement layer or the Lead layer. The ad platforms can count what they served and what was clicked, and they can count conversions: completed actions, such as form completions, that the platform claims as results of its ads. Each platform counts conversions by its own rules. Every platform sets its own attribution window, the period after an ad interaction during which the platform takes credit for a conversion. Most platforms also count view-through conversions, meaning conversions credited to an ad the person was shown but never clicked. And every platform counts alone: a buyer touched by a search ad, a social ad and a review-site listing is claimed in full by all three, so adding platform-reported conversions together produces a total comfortably larger than the number of real buyers.

The Pipeline and Revenue layers exist only in the CRM. A qualified opportunity is created by a salesperson, and a closed deal is recorded by finance. Neither event is visible to an ad platform. Every question senior management cares about, from cost per qualified opportunity to revenue per channel, can therefore only be answered from CRM data joined back to the media that ran.

Two methods connect CRM records back to the media that ran. The first is the click identifier: a tracking code added to the ad's landing page address and saved with the form completion, so the CRM can trace a lead back to the campaign that delivered the click. The second is account matching, used for creation channels and account-targeted media: comparing the companies appearing in the CRM against the accounts the media reached. Both methods lose records in predictable ways. B2B purchases are decided by a buying committee, the group of people at an account who approve a purchase together, and Gartner's research on the B2B buying journey puts that group at six to ten people. One member of the buying committee clicks the ad while a colleague completes the form weeks later, people move between devices, and tracking codes get stripped by security software. And even a complete record shows the order of events, not the cause: the CRM can show which touches came before the opportunity, and cannot show which touches produced the opportunity. The evidence behind that limitation is set out in why marketing attribution misleads.

Channels Work Together, and Should Be Measured Together

Everything above measures channels one at a time, and the channels do not work one at a time. Paid social and video build the memory that turns into branded searches. Paid search captures buyers that demand creation prepared. Retargeting maintains attention other channels earned. A buyer on the way to a purchase touches several channels, so reading each channel's numbers in isolation understates the channels that start the journey and overstates the channels that finish the journey.

Marketing mix modelling, often shortened to MMM and sometimes called media mix modelling, measures the channels together. MMM is a statistical method that relates spend by channel over time to business outcomes such as pipeline and revenue, using the natural variation in spend across months and markets to estimate how much each channel contributed, including the lagged effects of demand creation and the interactions between channels. MMM needs no tracking of individuals, so privacy changes and cookie loss do not weaken the method. MMM does need history: two or more years of spend and outcome data, meaningful variation in spend, and a budget large enough for the statistical estimates to settle. For companies with that history, MMM sits above the five layers as a cross-channel check on the whole plan. For smaller budgets, the practical substitutes are the CRM readings above plus deliberate tests, such as the holdout tests covered in Attribution vs Incrementality: The Difference That Moves Budgets. Whichever method fits, the principle stands: budget decisions should weigh the plan as a system of channels doing different jobs, never as a ranking of channels on one metric.

How to Build a Report Senior Management Can Trust

A trustworthy media report fits on one page. Five rules make the report trustworthy.

Show each channel at the layer that matches the channel's job. Paid search appears with cost per qualified opportunity in the quarter. Paid social and video appear with reach, frequency and engagement now, and with pipeline contribution from spend two or more quarters old. A report that holds every channel to a single cost per lead column looks tidier and misjudges half the plan.

Split leads by type. Demo requests, webinar registrations and content downloads each get their own row with their own conversion rate, never summed into one lead total, because a single summed total hides the intent differences between lead types that the worked example showed.

Never add platform totals together. Platform-reported conversions are for optimising inside a channel, comparing campaign A to campaign B under one set of counting rules. The CRM is the system of record for everything cross-channel.

State the demand creation lag on the page itself. One sentence does the work: pipeline attributed to creation channels reflects spend from two to four quarters ago. Without that sentence, every reader maps this quarter's creation spend to this quarter's pipeline, crediting the channel for old work and blaming the channel for results still maturing.

Build the report to invite the right questions. A well-built report leads senior management to ask which channels produce qualified opportunities cheaply, and whether creation spend from two quarters ago is now appearing at the Pipeline layer. A report built around cost per lead invites one question, why cost per lead moved, and the media team then spends the quarter answering a question from the wrong layer.

Paid Media Metrics FAQs

What are the five layers of paid media metrics?

Delivery (impressions, reach, frequency, viewability), engagement (clicks, view time, completions), lead (form completions, cost per lead), pipeline (qualified opportunities, cost per opportunity, pipeline value) and revenue (closed business, win rate, revenue per channel). Delivery proves the ads ran, engagement proves people noticed, leads prove people identified themselves, pipeline proves sales accepted opportunities, and revenue proves deals closed. Use the layer that matches the question being asked.

What is a good cost per lead in B2B?

No single figure answers the question, because cost per lead prices an event, and different lead events carry entirely different value. A $400 demo request converting to opportunity at 25 per cent is cheaper where cost matters, at the Pipeline layer, than a $30 download converting at under 1 per cent. Compare cost per lead only within a single lead type, and compare channels on cost per qualified opportunity with enough lag for leads to mature.

Why do platform-reported conversions never match the CRM?

Each platform counts by its own rules: its own attribution window, view-through credit for un-clicked ads, and modelled estimates where tracking is blocked. Every platform also claims any buyer the platform touched, so several platforms claim the same person. The CRM records opportunities and deals when sales creates them, under none of those rules. Use platform numbers to compare campaigns inside one channel, and use the CRM as the system of record for everything else.

Which metrics should demand creation channels be judged on?

Delivery and engagement in the quarter of the spend: reach into the target audience, frequency, viewability, view time and engagement rate. Pipeline and revenue two or more quarters later, joined through the CRM by account matching. Judging a creation channel on leads in the quarter of the spend is the most common reporting error in B2B, because creation channels reach buyers who are out of market and who have no reason to complete a form yet.

What should a monthly B2B media report include?

One page. Each channel reported at the layer matching the channel's job, leads split by type with conversion rates per type, cost per qualified opportunity by channel from CRM data, and a stated note that pipeline from creation channels reflects spend two to four quarters old. Platform-reported totals stay off the page, or appear clearly labelled, and no two platforms' conversion counts are ever added together.

Go Deeper

This post carries the judgement: which layers to use for each kind of channel, why the Lead layer misleads, and why cross-channel evaluation runs on the CRM. The method sits in the course Media and Measurement, which covers metric definitions layer by layer, CRM-based reporting design, lead-type taxonomies and the lag conventions for creation channels. The free module in B2B Marketing Fundamentals covers the strategy work that decides what the media should be asked to achieve in the first place. The difference between claiming credit and proving cause, which decides how far any joined report can be trusted, is the subject of Attribution vs Incrementality: The Difference That Moves Budgets.

Sources

• Les Binet and Peter Field, The Long and the Short of It, IPA, 2013: analysis of the IPA effectiveness databank showing that brand-building activity pays out over long periods while activation pays out quickly and fades, the basis for judging demand creation channels at the pipeline and revenue layers two or more quarters after the spend.
• Gartner, The B2B Buying Journey: buying groups of 6 to 10 decision makers, which is why one person's click and a colleague's form completion break the join between platform data and the CRM.
• Ehrenberg-Bass Institute for Marketing Science, 2021, How B2B Brands Grow: the finding that around 95 per cent of category buyers are out of market in any given period, which is why demand creation channels produce few leads in the quarter of the spend by design.