.png)
Imagine a medical test that tells a patient they have a serious illness, but is only right 5% of the time. Nineteen times out of 20, the result would be a false alarm, potentially leaving a patient frightened and facing unnecessary tests and treatment. It's difficult to imagine such a test being considered reliable.
Yet for roughly two decades, B2B marketing has tolerated something that raises a similar question about the reliability of its lead scoring. Long-running Gartner benchmark data puts the rate at which sales accepts MQLs at around 5–6%. While acceptance isn't the same as predictive accuracy, the figure raises an uncomfortable question: how useful is a qualification system if sales rejects the vast majority of the leads it identifies?
That's the starting point for Episode 3 of B2B Effectiveness: Evidence-Based Marketing Ideas for B2B Practitioners, in which Dale W. Harrison and Liam Moroney build on the previous episode's discussion of what an MQL should actually tell sales.
Their conclusion is that the issue runs deeper than the signals being used or how they're weighted. The conventional points-based approach has fundamental limitations that prevent it from delivering the predictive value marketers expect.
Why conventional lead scoring keeps falling short
Over the years, the industry has introduced one alternative after another: Marketing Qualified Accounts, Marketing Qualified Opportunities, AI-qualified leads and other attempts to improve on the MQL. Yet many of these approaches leave the underlying scoring mechanism intact.
The mechanism is familiar. Behavioral or demographic signals are assigned numerical values and added together. Visiting an About page might be worth one point; downloading an eBook might be worth ten. But where did those numbers come from? In most cases, they weren't derived from any meaningful relationship between the behavior and eventual commercial outcomes. They were chosen and adjusted until the resulting scores seemed useful.
AI-driven scoring is often presented as a more sophisticated answer because it can process far more signals than a human-defined model. But if the underlying logic still treats observed activity as evidence that can simply be accumulated, adding more data and sophisticated mathematics doesn't necessarily make its predictions more reliable.
Data is not information
Dale illustrates the distinction with a simple example. Ask someone a question and they give you an answer. Ask the same question ten times and receive the same answer each time. You now have ten data points, but you haven't acquired ten times as much information.
That matters in lead scoring. If someone downloads ten articles from your site, the activity may tell you something useful about their interests and fit with your target audience. But downloading 100 articles doesn't make you ten times more certain they're a good prospect. At some point, additional activity stops telling you anything new.
A points-based model can struggle to recognize that distinction. If every download continues to add points, repeated behavior is treated as fresh evidence even after its informational value has largely been exhausted.
The importance of distinguishing wins from losses
This becomes especially important when models are built from historical sales data. A common approach is to examine accounts that became customers, identify their shared characteristics and use those patterns to predict future buyers.
But a characteristic shared by your winners only matters if it distinguishes them from your losers.
Suppose 31% of your closed-won accounts visited your LinkedIn company page. That might look like a useful signal. But if 31% of your closed-lost accounts visited the page too, the behavior tells you nothing about the likelihood of winning. It's equally common to both outcomes.
Useful patterns need to occur at materially different rates between wins and losses, ideally consistently enough to improve predictions beyond what you'd achieve without the model.
That sounds obvious, but many lead scoring systems rely on correlations that haven't been properly tested against the outcomes they're supposed to predict.
What traditional scoring fails to capture
Even a model that identifies patterns in historical data needs to be tested against what happens in practice. That means examining both the leads it identifies as promising and those it overlooks.
Measuring what you're missing
Most teams focus on false positives: leads that the model scores highly but that never become customers. The other side is just as important. What happens to the leads the model scores poorly that later become genuine opportunities?
Answering that question requires closed-loop analysis. Dale describes tracking leads in real time, then returning to those records months later to see which entered an active, sales-qualified process, regardless of their original score.
That analysis can reveal some uncomfortable patterns. One is what Dale calls “click monkeys”: people who click on everything, download everything and appear highly engaged, yet never become customers. Their digital behavior can look remarkably similar to that of a serious buyer.
The reverse also happens. Some strong buyers leave almost no meaningful digital footprint before entering a sales process. They may have been referred by a trusted colleague, used the product at a previous company or inherited an existing supplier relationship.
Much of this activity happens outside the channels marketers can observe and score, often described as dark social. As a result, some of the signals that most influence a purchase may never appear in the scoring model.
The buyer is scoring you too
There's another important consideration: while companies spend considerable effort scoring buyers, buyers are evaluating companies at the same time.
A product page filled with generic marketing language might carry little weight with a buyer. A recommendation from a trusted colleague, or previous experience using the product, could carry considerably more. Yet these influences are often difficult for marketers to observe.
This is where the relationship between buyer and seller becomes important. Two businesses can sell similar products and achieve very different close rates because their experience, reputation and capabilities are concentrated in different markets. One might have expertise in healthcare and government procurement, while another is better positioned to win industrial and manufacturing customers.
Those differences affect the likelihood of winning, but a conventional engagement score is unlikely to capture them. Qualification needs to consider the fit between a particular buyer and seller, rather than simply how active the buyer has been.
A useful scoring system needs to prove its value
Strip away the framework names – MEDDIC, BANT and whatever comes next – and the underlying test for any qualification system is relatively simple: does it predict outcomes better than you could without it?
If you know nothing about a lead, you can only work from the average likelihood of winning. A useful scoring system needs to improve on that baseline by identifying patterns that reliably distinguish stronger opportunities from weaker ones.
That means testing predictions against real commercial outcomes, including both the leads the model prioritizes and those it overlooks.
The persistently low MQL acceptance rates reported by Gartner suggest that much of the industry still has work to do. We've built increasingly elaborate systems, added more data, introduced AI and given them more sophisticated names. But greater complexity doesn't necessarily mean better qualification.
The real test is whether the model helps sales make better decisions about where to invest its time.
In Episode 4, Rising Markets Break the 95:5 Rule, Dale and Liam explore another factor that can change those odds: the market itself. How does the proportion of buyers actively looking to purchase change when a category is growing or shrinking, and what does that mean for qualification?
