AMA: Counterpart Marketing Lead, Erika Barbosa on Funnel Science

Accurate pipeline forecasting requires moving away from static stage-weighted models that apply a single conversion rate to every deal in a given stage. The problem with that approach is it treats a $200K enterprise deal sitting in stage 3 for 45 days the same as a $20K mid-market deal at the same stage and age. Those are completely different probability profiles and conflating them is where forecast accuracy breaks down.The input I find most underused is leading indicator data from earlier in t ...Read More

Governance without enforcement is just a style guide nobody reads. The only UTM and event schema frameworks that actually hold up across channels, regions, and teams are the ones baked into the tooling itself, not documented in a doc somewhere. Here is how I have approached this: Centralize taxonomy ownership in one place: Someone has to own the master taxonomy and have the authority to say no to exceptions. In practice this is usually a marketing ops or data engineering function, but the key is ...Read More

Cohort analysis only becomes useful for controlling seasonality and aging when you are consistent about which anchor date you use and why. First-touch month is the cut I rely on most heavily because it controls for the market conditions that existed when a prospect entered the funnel. Comparing conversion rates without this anchor conflates funnel performance with external factors like budget cycles, competitive activity, and macro conditions. A Q4 cohort will almost always behave differently th ...Read More

Most teams discover channel saturation after the fact, when CPLs have already climbed and pipeline has already softened. The goal is to build systems that surface those signals early enough to act on them.The earliest reliable indicator I watch is CPL trending up while volume stays flat. You do not need a sophisticated response curve model to catch this, consistent weekly data by channel that separates volume from efficiency will show it. When a channel's rolling 4-week CPL exceeds its 90-day ba ...Read More

I default to expected value scoring over ICE or RICE because it forces you to think in revenue terms. ICE and RICE are fine for early-stage prioritization, but they break down when you need to defend resource allocation to a CFO. Here is how I build the framework in practice: Expected value as the foundation: The formula is simple, probability of success multiplied by estimated revenue impact. The hard part is being honest about your probability estimates. I pull those from historical experiment ...Read More

Product signals become meaningful funnel inputs only when you define thresholds that correlate with downstream revenue, not just activity. Many teams make the mistake of calling any active user a PQL, but the definition needs to be grounded in data. Here is how I approach building this out: Start with outcome-based PQL definitions: Pull your closed-won deals and work backward. What product behaviors did those accounts exhibit in the 30, 60, and 90 days before opportunity creation? Frequency of u ...Read More

Meetings booked is the single strongest predictor of win rate in every ABM model I've built, but the relationship only holds when you look at meeting quality, not just volume. Here's what I've found actually moves the needle: Meetings booked (ABM vs. non-ABM accounts): The delta here is striking. In most models I've run, ABM accounts that reached a first meeting converted to pipeline at 2-3x the rate of non-ABM accounts. That gap alone justifies the program investment. Multi-threaded engagement ...Read More