L'engagement est important si vous savez comment le mesurer grâce aux segments ICP.

Every company tracks engagement like website visits, email opens, content downloads, and event attendance. They score it and report it, but most times they don’t know what that data predicts.

Engagement matters when you measure it forward. The problem is that organizations measure it backwards, meaning they import generic definitions of “account engaged” and assume that fits their business. 

There are three critical steps to doing engagement right: understanding which customer segments drive revenue, discovering which signals show those segments are in-market, and defining what “engagement” means for their specific business.

Without that foundation, engagement metrics are vanity.

When Engagement Metrics Are Misleading

General definitions of an engaged account are: three or more website visits in 30 days. Opened two emails. Downloaded a whitepaper. Attended a webinar. Clicked an ad on LinkedIn. Visited the pricing page. Watched a demo video.

These get rolled into an “engagement score” that Marketing usually reports to leadership as part of their goals, but it is common that those numbers don’t materialize into pipeline. 

These metrics treat all activity as equal signals of buying intent. But activity doesn’t equal intent.

People visit your website for a lot of reasons that have nothing to do with buying. Same with email opens, content downloads, and event attendance. Without context, you can’t distinguish between casual browsing and serious buying behavior.

When engagement metrics don’t distinguish signal from noise, Sales wastes time chasing “engaged” accounts that aren’t in-market, and RevOps can’t forecast accurately because “engagement” doesn’t correlate with pipeline

The result: leadership decides based on vanity metrics, unpredictable revenue, and a GTM engine running wrong data.

The Engagement Metrics Problem Deconstructed

Layer 1: No Buying Journey Context

The same signal has different meanings during the buying journey.

For example, a website visit. During the Unengaged stage, it could be casual browsing, research, or awareness. In the Qualified stage, that visit likely signals active evaluation, like serious research. Website visits in the Active Pipeline stage can show urgency or progress; they’re looking over materials before a decision meeting.

Generic engagement scoring treats all three identically, without taking into consideration that a website visit from an account you’ve never heard of is not the same as a website visit from an account in an active pipeline.

I like how ForgeX‘s framework categorizes accounts by buying stages using signal frequency: Unengaged (few signals), Engaged (some signals), MQA (moderate signals), Active Pipeline (many signals), and Closed-Won (very many signals); it helps to map signals to these stages.

Layer 2: No Signal Pattern Discovery

Defaulting on analyzing closed-won deals to understand what signals appeared and what signals were absent and looking for patterns usually leads to a situation where your are mapping signals to stages but still don’t know which signals predict revenue.

Assuming that “more engagement equals better” without analyzing deeper will cause measuring correlation instead of causation. Just because engaged accounts sometimes win doesn’t mean the engagement caused the win.

Layer 3: No ICP Segment Understanding

Different ICP segments show different signal patterns.

An example from AlignICP illustrates this perfectly. A seventy-million-dollar B2B SaaS company asked its GTM teams “Which industry vertical performed best last year?”

Sales and Marketing answered: Entertainment and Travel. Customer Success and Product answered: Retail.

Both were right (for their metrics).

Entertainment and Travel showed the highest win rates and highest average sales prices. These accounts generated lots of “engagement” signals early in the sales process. Sales loved them because they closed fast and delivered big deals. But these same accounts had high churn rates, limited expansion potential, and poor retention. They looked like wins in the moment but bled revenue.

Retail showed lower win rates and was harder to engage early. Sales found them frustrating because they took longer to close and required more effort. But these accounts had the best retention rates, highest Net Promoter Scores, and strong expansion revenue. Customer Success loved them because they stayed, grew, and became advocates.

When you plot customer segments across these two dimensions, you get four quadrants. Flywheel accounts are easy to acquire and retain strongly. These are your true ICP. False Positive accounts show that they’re easy to acquire but churn quickly. This is the engagement trap. High Potential accounts are harder to acquire but retain strongly once you land them. These segments need better messaging. You have to avoid accounts that are hard to acquire and that have high churn. 

If you’re measuring “engagement” without ICP segmentation, all these patterns will mix together. Without ICP segmentation, your engagement metrics are averaging across all four quadrants.

Fixing Engagement Metrics

By following the right sequence.

Step 1: Discover Your ICP Segments

Start with your customer database. 

For acquisition, look at which segments have the highest win rates, which close fastest based on sales velocity, and which deliver the highest average sales price. For retention, look at which segments have the highest customer lifetime value with high net revenue retention.

Then segment your customer base by plotting customers across the acquisition and retention matrix. Flywheel accounts with high acquisition and high retention are your true ICP.

Define targetable attributes for each segment like industry and sub-industry, company size based on revenue or employees, geography, tech stack and techno graphic signals, and primary use case or pain point.

Step 2: Discover Signals That Indicate Your ICP Segments Are In-Market

With ICP segments defined, audit your closed-won and closed-lost data to find signal patterns.

For closed-won deals, ask: What signals appeared consistently? When did they appear in the buying journey? Which GTM touchpoints generated those signals?

For closed-lost deals, ask: What signals were absent? Where did engagement drop off? Which touchpoints failed to generate signals?

The ForgeX Signal Pattern Analysis framework breaks signals into two categories. GTM Engagement Touchpoints are first-party signals like direct inquiries, meetings, events, content downloads, website visits, and email engagement. Buying Intent Signals are second- and third-party signals like job changes, keyword intent spikes, tech adoption patterns, and hiring signals.

The foundation for meaningful engagement metrics is a documented map of which signals predict revenue for each ICP segment, which combinations of signals matter most, and which signals are noise.

Step 3: Define What “Engagement” Means for Your Company

Now you are ready to build engagement scoring that predicts pipeline.

Weight signals by three dimensions.

First, by ICP segment. For Flywheel accounts, weight signals that predict both acquisition and retention. For False Positive accounts, de-prioritize or flag signals for caution and for High Potential accounts, weight signals that show messaging is landing.

Second, by buying journey stage. At Unengaged, prioritize early awareness signals like ad clicks and website visits. At Engaged, look for multi-touch engagement and content consumption. At Qualified, prioritize high-intent signals like pricing page visits, demo requests, and buying group engagement. At Active Pipeline, weight direct outreach responses and multi-stakeholder meetings.

Third, by historical correlation with revenue. Signals that appeared in eighty percent or more of closed-won deals get high weight. Signals present in both won and lost deals get lower weight. Signals absent from high-LTV accounts should be de-prioritized entirely.

When your RevOps team operationalizes this in your CRM, you will have engagement metrics that predict pipeline, segment by ICP, and adapt as signal patterns evolve.

Why This Is Hard

Because cross-functional alignment is the first barrier. ICP discovery requires Sales, Marketing, Customer Success, RevOps, and Finance to agree on definitions and share data.

Messy CRM data is the second barrier. Signal pattern analysis requires clean, structured data. 

Analysis complexity is the third barrier. Building acquisition and retention segments requires statistical analysis.

Short-term thinking compounds the problem. Leadership wants results this quarter. “Engaged accounts” is an easy metric to report. Doing ICP analysis, signal discovery, and custom scoring takes longer and the shortcut is to use generic scoring, and set goals with vanity metrics.

Measuring It with Context

Engagement isn’t the problem. Measuring it without context is.

If you don’t know which customer segments drive revenue, which signals indicate they’re in-market, and what “engagement” means at each stage of their journey you’re measuring noise.

Fix the foundation first. Discover your ICP segments. Map the signals that matter. Define engagement based on your data.

Engagement Metrics in ABM Takeaways