Enhancing the Waterfall: Prospects, MQLs and the Steps in Between

The demand waterfall, popularised by the research firm SiriusDecisions (now part of Forrester), is one of the most useful pictures in B2B marketing. It shows that marketing and sales are one process, and that both teams must agree on a single definition of a lead before it is handed over.

The waterfall has been revised several times since this post was first written, and in recent years has evolved into models built around buying groups and opportunities rather than individual leads. But two practical problems it raised in 2010 still trip teams up, and both are about how you turn broad stages into a programme you can actually run.

Problem 1: the top of the funnel is missing

The classic waterfall starts with inquiries, the people who have already raised their hand. Demand generation starts earlier, with the pool of prospects you need to reach to get those inquiries. Without that number you cannot do reverse funnel maths, and without reverse funnel maths you cannot justify a budget with a formula your CFO will accept.

Adding a Prospect stage lets you work backwards from closed deals to opportunities, leads, respondents and finally the number of people you need to touch through search, social, events and lists. The buyer-targeting context post shows the calculation.

Problem 2: "MQL" hides most of the work

A marketing-qualified lead is usually described as one judged worthy of hand-off to sales, based on nurturing, a demographic profile and further interactions. True, but that is like telling a cook that a good meal is made of ingredients that taste good. It names the outcome without the recipe.

Nurturing and scoring are where most programmes struggle, so break the MQL stage into smaller steps, each with its own entry criteria and content:

Scoring: fit plus behaviour

Most scoring models combine two dimensions. Fit (sometimes called demographic or firmographic score) asks whether this is the kind of person and company that buys from you: industry, size, role, region. Behaviour asks how engaged they are: content consumed, pages visited, events attended, replies. A lead becomes an MQL when both are high enough. High behaviour with poor fit is a student or a competitor; high fit with no behaviour is a name on a list.

A score is a probability, so treat it like one

The weakness of many scoring models is that the points are arbitrary. Forty points for a webinar, ten for an email click: why? A better approach borrows from the gambling world, where every number is a probability and every decision is an expected-value calculation. A blackjack player who knows basic strategy is not guessing; they are choosing the action with the best expected outcome given the cards. A scoring model should work the same way: points should reflect how much each signal actually raises the likelihood that a lead becomes an opportunity, measured on your own historical data.

We work through that method, with a roulette wheel and some simple arithmetic, in the probability mindset in lead scoring. It turns the MQL threshold from a negotiation into a calculation: pass the lead when the expected value of a sales call is higher than its cost.

Making the waterfall run

• Write entry and exit criteria for every stage and publish them to both teams.
• Track stage-to-stage conversion and time in stage every month.
• Recycle, do not discard: rejected leads go back to nurturing with a reason code. See the layered nurturing model.
• Agree a service level for follow-up of sales-ready leads, as described in closing the sales-marketing gap.