Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Content Marketing › Content Distribution

Measuring Content Performance

How to measure content: metrics by stage, engagement versus outcomes, assisted conversions and content attribution, and grouping pages for reporting.
Guideintermediate10 min readUpdated

What content measurement must separate

Goal setting and the choice of KPI belong to the content goals and KPIs guide; this guide starts where that one ends, with the mechanics of measuring the goals once set. The first mechanical problem is that "did the content work" is three questions wearing one coat.

The first is discovery: did the people it was written for find it? That is a function of distribution and search, measured in impressions, visits and their sources, and it is the stage most reports stop at. The second is engagement: once there, did they read, watch, scroll, return or act within the piece? That is a function of the writing and the match between promise and content. The third is contribution: did the piece play a part in an outcome the organisation cares about, such as a sign-up, an enquiry, a purchase or a retained customer? That is a function of the piece's place in the journey and the calls to action it carries.

The three move independently. A piece can be found by thousands and engage none, engage deeply and contribute nothing, or contribute heavily from a small, well-targeted readership. Blending them into one score hides which stage failed and therefore which part of the workflow needs attention. Measurement that keeps them apart points at the brief, the writing, the distribution or the call to action, and a monthly review can act on that.

Metrics by stage

Each stage of the reader's journey has metrics that fit it. The table pairs stages with the metrics most teams can collect from a standard analytics setup and the question each answers. The stage labels follow the funnel used across this site; the measurement-plan framework is the exercise of choosing one row per piece before it is published.

StageWhat the piece is forDiscovery metricsEngagement metricsContribution metrics
AwarenessBeing found by people who do not know the organisationOrganic impressions and clicks by query; new visitors; referred visits by channelScroll depth; engaged sessions; time on page relative to length; returning readersSubscriptions and follows started from the page; assisted conversions much later
ConsiderationHelping people compare and understandVisits from search, email and referrals; share of returning visitorsPages per session within the pillar; downloads; video completion; next-page choiceLeads, demo requests, template downloads; assisted conversions
DecisionRemoving the last doubtsVisits from product pages and sales emailsComparison and pricing views after reading; chat or contact useDirect conversions; conversion rate (CVR) of readers versus non-readers
RetentionKeeping customers successfulVisits from in-product links and support; logged-in readershipReturn visits; help-content completionFewer support contacts; renewals among readers, where the data joins

Discovery metrics for search-led pieces are the subject of the SEO KPIs and reporting guide; the content report borrows them rather than recomputing them. Engagement metrics depend on the analytics configuration, and GA4 in particular defines engaged sessions and engagement rate in its own documented way, so the definition the report uses should be stated once and kept.

Engagement vs outcomes

Engagement metrics are the ones a team has most of and trusts most, and they are the ones most often misread. Time on page is a function of length and of whether the reader left the tab open; scroll depth is a function of layout; page views are a function of distribution, not quality. None of them is useless, but each measures the reading, not the result, and a report that leads with them invites the question "so what".

Outcomes are what the organisation would pay for: the subscription, the lead, the sale, the retained customer. For most content they are rare, delayed and indirect, which is why teams retreat to engagement. The discipline is to report both, side by side, for every piece or group, and to refuse to present either alone. Engagement explains outcomes: a piece with strong outcomes and weak engagement is probably converting a small, precise audience and could reach more of it; a piece with strong engagement and no outcomes is probably missing a call to action or sitting at the wrong stage. Outcomes justify engagement: an awareness piece whose readers subscribe at a reasonable rate has earned its page views.

Where the outcome happens on the same site, standard analytics can relate the two through key events, conversion counts and the pages that preceded them. Where the outcome happens elsewhere — in a sales system, a product, a store — the join requires an identifier carried from the content to the outcome, and the measurement plan has to say who owns that join before anyone reports on it.

Content attribution and assisted conversions

Attribution for content asks which pieces appeared in the journeys that ended in an outcome, and how much credit each should get. Analytics tools answer the first part by recording the pages and sessions that preceded a conversion and the second by applying a model. Google Analytics documents several attribution settings and reports that show conversions by the channels and touchpoints that contributed; which models are available has changed over time, so the current help documentation is the reference, and this guide avoids stating which models exist today.

Two reports matter most for content. The first is the set of conversions a page assisted: journeys in which the page was visited before the converting session, whether or not it was the last touch. For content, assisted figures are usually larger than last-touch figures, because people read before they act and act somewhere else. The second is a path analysis: the pages and sources that commonly precede a conversion, read in order. It shows which pieces sit early in journeys, which sit late, and which never appear, and it is the closest thing to evidence that a piece is doing the job its stage implied.

Every attribution figure is a modelled estimate. It depends on the model's rules, on how much of the journey was tracked, on consent choices that remove sessions from the data, and on cross-device journeys the tool cannot stitch. Reporting a content-attributed revenue figure as cash earned invites a dispute that cannot be won; reporting it as "modelled contribution under the named model, with these known gaps" invites a decision. The attribution section covers the models and their limits in depth. Campaign parameters (UTM) on every distribution link, following the naming conventions and UTMs guide, are what make the channel part of any journey readable at all; without them, distribution collapses into "direct" and attribution has nothing to work with. Video published on a platform is harder still, because most viewers never click, and the guide to measuring YouTube's contribution to the business sets out how to report an organic channel as a range rather than a last-click figure. For a programme that exists to produce enquiries, the guide to content for lead generation uses assisted pipeline in the same way, alongside the qualified leads each asset sources directly.

Illustrative example — not a real result

A quarter of content attribution for a software company

A project-management software company publishes 14 guides in a quarter and tracks every distribution link with campaign parameters. Last-touch reporting credits the guides with 38 trial sign-ups, which the finance team calls disappointing against the 290 sign-ups the quarter produced. The assisted-conversions report tells a different story: guides appeared in the journeys of 131 sign-ups, and the path analysis shows a consistent pattern of a guide visit from search, a return by email a week or two later, and a sign-up from the pricing page. Grouped by pillar, the "team workflows" pillar assisted 84 of those sign-ups from 5 guides, while the "integrations" pillar assisted 9 from 6 guides. Under the data-driven model in the analytics tool, content is credited with a modelled 23% of the quarter's trial sign-ups. The report states the model, notes that 31% of sessions declined analytics consent and are missing from the paths, and recommends two more "team workflows" guides and a review of the integrations pillar's calls to action.

Grouping content for reporting

A content report at the URL level is unreadable past a few dozen pages and says nothing about strategy. Grouping fixes both. The three groupings most teams need are by pillar (which theme the piece belongs to), by type (guide, tutorial, story, template, video) and by stage (the row of the metrics table it was written for). Pillar grouping answers whether the strategy's themes are working; type grouping answers which formats earn their production cost; stage grouping answers whether the funnel has gaps.

Analytics tools support grouping in different ways. Google Analytics documents a content group dimension, set by sending a parameter with page views, and the reports and dimensions guide describes how dimensions like it appear in standard and custom reports; where that is unavailable, a lookup table joining URL to pillar, type and stage, maintained alongside the content inventory, does the same job in a reporting tool. The groupings must be assigned when a piece is published, not reconstructed at reporting time, because retrofitting them across hundreds of pages is the task that never gets done. The content inventory and audit template keeps each page's pillar and stage in the same row as its performance, so the grouping is already there when an audit reads the library.

Grouped reports also make comparison fair. A tutorial should be compared with other tutorials, an awareness piece with other awareness pieces, and this quarter's pillar with the same pillar last quarter rather than with a different pillar's best month. The KPI trees guide shows how grouped content metrics roll up into the content KPI and from there into the marketing figures the organisation reports on.

The monthly content review

Measurement is only worth doing if a meeting ends in a decision, and the monthly content review is that meeting. It takes the grouped report and asks a fixed set of questions.

  • Which groups are up or down against their own history, and why? Changes are traced to a cause: a distribution change, a search ranking shift, a seasonal pattern, a broken tag. Unexplained changes are investigated before anyone celebrates or panics.
  • Which pieces published this quarter have not been found? Discovery failures point at distribution and on-page work, not at the writing; the piece may need a second distribution sequence or an internal-link pass.
  • Which pieces are found and read but contribute nothing? Engagement without outcomes points at the call to action or the stage; the piece may need a different next step or to be linked from a decision-stage page.
  • Which pieces contribute far more than their size? Over-performers deserve derivatives, refreshes and promotion, and their pattern — topic, format, angle — informs the next quarter's briefs. When that promotion is paid, the tutorial on promoting content with paid media tags each promoted derivative and reads its engaged visits against what the owned channels delivered.
  • What should be refreshed, merged or retired? Decaying pieces are flagged for the governance routine.
  • What does the next month's calendar change as a result? The review writes the changes into the calendar and the briefs, which is the only proof that measurement had an effect.

The review is short when the report is grouped and the questions are fixed. The distribution hub sets the routine for each piece; the review is where the routine's results are read. To check whether this reporting, and the programme around it, is doing its job, the content marketing programme audit scorecard scores performance reading alongside goals, standards, distribution and upkeep.

When articles draw steady readers and the outcomes stay flat, the diagnosis for blog traffic that is not converting separates a missing next step, the wrong readers and assists hidden by last click.

Whether refreshing decaying articles earns back its editorial hours can be tested with the content refresh experiment, which compares refreshed articles with an untouched control group.

Common mistakes

  • Leading with page views. Views measure distribution, not quality or outcome, and a report that opens with them gets asked "so what" by everyone who matters.
  • One blended score. A single content score hides which stage failed and therefore which part of the workflow to fix.
  • Last-touch only. Content is read before action and rarely gets the last click; assisted figures and path reports are where its contribution shows.
  • Attribution figures reported as fact. Modelled contribution presented as revenue earned invites a dispute with finance that the content team loses.
  • Groupings assigned at reporting time. Retrofitting pillar, type and stage across hundreds of URLs never gets finished; assign them at publication.
  • No tracking on distribution links. Without campaign parameters every channel becomes "direct", and the distribution plan cannot be evaluated.
  • Reviews with no decisions. A monthly report that changes nothing in the calendar is a cost; the review ends when the briefs change.

Frequently asked questions

What is the single best metric for content?

There is none, and looking for one is how teams end up reporting page views. Content works at several stages and a metric that measures one stage says nothing about the others. The workable answer is one metric per stage the piece was written for, with the outcome metric reported alongside the engagement metric so neither can be mistaken for the other.

How long after publication should content be judged?

It depends on the channel that carries it. Email and social distribution show their result within days; organic search often takes months to settle, and evergreen pieces keep accruing for years. Judge distribution-led pieces at the end of the release sequence, search-led pieces after a full quarter, and review every piece again at the annual audit before deciding to refresh, merge or retire it.

Can content be credited with revenue?

Partly, and with care. Analytics tools can show which content pages appeared in the journeys that ended in a conversion, and attribution models distribute credit across those touchpoints according to rules the tool documents. The figures are estimates whose meaning depends on the model, tracking coverage and consent, so report them as a modelled contribution with the model named, never as cash the content earned.

Terms used on this page

Frameworks

Where this page is used

Part of these learning paths

Sources

  1. Google Analytics Help — Google Analytics Help

Related