The Demand Waterfall — Infographic

The outcome you should expect

Teams adopt the Demand Waterfall expecting a prettier funnel chart. That is not the payoff. The payoff is that arguments about lead quality stop being opinion-based within about one quarter of honest measurement, because every claim now resolves to a specific stage-to-stage conversion rate that both marketing and sales agreed to define in advance.

Concretely, here is what a working waterfall changes in the first 90 days. Before, a sales leader says "marketing leads are garbage" and a marketing leader says "sales never works them." After, both parties look at the same number — say, MQL-to-SAL running at 14% against an agreed target of 25% — and the conversation moves to *why* that specific handoff is failing. That is a materially different meeting. It is also a meeting you can only have if the stage definitions were written down before the numbers were bad, because definitions written after the fact get bent to fit whoever is losing the argument.

The second outcome is forecast credibility. Once you know that opportunities convert to closed-won at roughly 22% with a 45-day median cycle, you can back-calculate how many SQLs the top of the funnel must produce this month to hit a number two quarters out. Without stage-level rates you are forecasting off gut feel and a weighted pipeline number that the CRM computed from probabilities nobody calibrated.

The third outcome is budget defensibility. When a CFO asks what the marketing spend bought, "we generated 4,200 inquiries" is a weak answer. "We generated 4,200 inquiries, which produced 780 MQLs, 210 SALs, 96 opportunities, and 21 closed-won deals" is a chain of custody from spend to revenue. Every stage in that chain has a rate you can attack when it underperforms.

What you should *not* expect: the waterfall does not create demand, does not fix a weak product-market fit, and does not resolve territory or comp disputes. It is a measurement instrument. If the underlying demand is thin, the waterfall will show you thin demand with excellent precision and nothing more. Teams that expect the model itself to lift conversion are consistently disappointed; teams that treat it as a diagnostic and then act on the diagnosis are not.

Finally, expect the first version to be wrong. The initial stage definitions will be too loose in one place and too strict in another, and you will find out only after two or three months of data. Plan for a definition revision at the end of quarter one. That revision is a sign the instrument is working, not a sign the project failed.

What drives that outcome

Three mechanisms drive whether a waterfall produces real decisions or becomes decorative: stage definitions with enforceable entry and exit criteria, a scoring model that predicts rather than flatters, and a disposition loop that returns stalled records to nurture instead of letting them rot.

Stage definitions. Each stage needs an entry rule and an exit rule that a second person could apply identically. Inquiry entry is usually any form fill, webinar registration, or content download — but you must decide explicitly whether bot submissions and free-email-domain records count. Most teams exclude them by requiring a resolvable corporate domain. MQL entry should never be a single score threshold; combine a behavioral signal (repeat visits to pricing, a demo request, a competitive comparison page view) with a fit signal (headcount band, industry, seniority in title). SAL means a human on the sales side explicitly accepted the record within a defined window — commonly 24 business hours — after reviewing it. Automating acceptance destroys the entire purpose of the stage, because acceptance is the only place where sales publicly takes on ownership. SQL means a discovery conversation happened and budget, authority, need, and timing were confirmed or explicitly disqualified. Opportunity means a CRM record with a dollar value and a close date, created only after the buyer agreed to a concrete next step.

Scoring that predicts. The engine feeding MQL is where most implementations go wrong. Points for opening an email are close to worthless — the signal-to-noise ratio is terrible and it inflates volume without moving revenue. Weight actions by their observed correlation with reaching SQL: pricing page visits, demo requests, and comparison-page views earn heavy weight; generic top-of-funnel downloads earn very little. Add negative scoring — unsubscribes, student or intern titles, free email domains, and long inactivity should all subtract — because without it every list eventually drifts upward and the threshold stops meaning anything. Apply time decay so a record that scored high six months ago is not treated the same as one that scored high last week; a monthly decay of roughly 10% of accumulated behavioral points is a common starting point.

The disposition loop. Every stage needs a maximum dwell time and an automatic outcome when it expires. A record sitting in SQL for 60 days with no logged activity is not an SQL; it is a fiction inflating your coverage ratio. Recycle it to nurture, mark it disqualified with a reason, or reopen it deliberately. Without this loop, waterfall volume only ever goes up, which makes the chart look healthy while pipeline quality quietly degrades.

The recycle edge back to the top matters more than it looks. In most B2B motions the majority of records that eventually buy were disqualified at least once for timing. If your waterfall has no return path, you are permanently discarding your highest-intent future demand and paying to reacquire it later.

Benchmarks and realistic ranges

Published conversion benchmarks are the single most misused artifact in demand generation, because they get quoted without the context that determines them: deal size, sales cycle length, motion type, and how strictly the reporting team defined each stage. A team with a strict MQL definition will show a low inquiry-to-MQL rate and a high MQL-to-SAL rate; a team with a loose definition shows the mirror image. Comparing those two teams on any single rate is meaningless.

Use ranges as sanity checks, not targets. Broadly, in mid-market B2B software with deal sizes in the low tens of thousands, teams commonly see inquiry-to-MQL somewhere in the mid-teens to mid-twenties percent, MQL-to-SAL in the twenties to low thirties, SAL-to-SQL considerably higher because a human already vetted the record, SQL-to-opportunity in the thirties to fifties, and opportunity-to-closed-won in the twenties to low thirties. Every one of those spans is wide because the underlying definitions vary enormously between companies.

Deal size shifts the whole curve. As average contract value climbs, buying committees grow, evaluation cycles lengthen, and per-stage conversion rates fall — but the revenue per surviving record rises fast enough to compensate. A team closing 25% of opportunities at a small deal size and a team closing 12% at ten times the deal size can be equally healthy businesses. Judging the second team against the first team's rate would push them toward chasing volume that their sales capacity cannot absorb.

The far more useful benchmark is your own trailing twelve months. Pull each stage-to-stage rate by month, plot it, and establish your baseline and your normal variance. A rate that swings between 18% and 26% month to month has a normal band of roughly eight points; a single month at 21% is not a signal. A month at 11% is. This is the difference between running the waterfall as a control chart versus reacting to noise, and it is the most common failure of monthly demand reviews.

Segment before you conclude anything. Aggregate rates hide the interesting variance. Split by acquisition channel, by segment, and by product line. It is routine for one channel to convert several times better than another at the same stage — for example, records sourced from a partner referral or a high-intent search term converting far better than a broad content syndication list. The aggregate rate is the volume-weighted blend of those, so an aggregate decline often means nothing more than that the mix shifted toward a cheaper channel. If you optimize on the aggregate you will make exactly the wrong call.

Volume matters as much as rate, and leads it in time. A 20% drop in monthly MQL volume with rates unchanged is a top-of-funnel problem that will not show up in bookings for one full sales cycle — often three to six months. By the time revenue reflects it, the quarter is already lost. Watching stage volume as a leading indicator is where the waterfall earns its keep, and it is precisely what an Infographic rendering makes legible to executives who will never open the underlying report.

Finally, track velocity alongside rate. Median days-in-stage is a separate diagnostic. Two teams with identical conversion rates but a 30-day versus 75-day median opportunity stage have completely different cash-flow profiles and require different pipeline coverage ratios. If your rates are stable and your revenue is missing, velocity is usually the culprit, and no rate-only view of the waterfall will surface it.

Risks, edge cases, and failure modes

Definition drift. The most common failure is silent. Someone adjusts a scoring rule, an ops admin changes a lifecycle stage automation, a new campaign type gets mapped to the wrong stage — and six weeks later the trend line breaks and nobody can explain it. Mitigation: version your stage definitions in a shared document with a changelog, and annotate the waterfall chart with the date of every definition change. When you see a step change in a rate, check the changelog before you investigate behavior.

Automated acceptance. If SAL is granted by a workflow rather than a person, the stage measures nothing. It becomes a copy of MQL with a different label, and you lose the only checkpoint where sales publicly commits to working a record. Teams do this because manual review feels like friction. That friction is the product.

Double counting at the account level. Six people from the same company download the same asset and you now have six MQLs and one buying opportunity. In account-based motions this inflates top-of-funnel volume dramatically while opportunity counts stay flat, producing a conversion rate that looks catastrophic and is actually an artifact. Decide early whether the waterfall counts people or accounts, and if you need both, build two views and never mix them on one chart.

Attribution collisions. A record that enters via paid search, goes cold, returns nine months later via a webinar, and closes will be claimed by both channels unless you have an explicit rule. Pick a convention — first touch, last touch before opportunity creation, or a documented multi-touch model — and apply it consistently. The convention matters less than the consistency; switching conventions mid-year makes year-over-year comparison impossible.

Recycled-lead accounting. When a disqualified record re-enters, does it count as a new inquiry? If yes, your inquiry volume inflates and your inquiry-to-MQL rate degrades even though nothing changed. If no, you undercount genuine re-engagement. The workable compromise is to count re-entries separately and report them as a distinct line so the base rate stays clean.

Gaming the handoff. Where compensation is tied to a stage count, that stage count becomes unreliable. If SDRs are paid on meetings booked, meetings get booked with unqualified buyers. If marketing is paid on MQL volume, the MQL threshold quietly lowers. Compensate on stages as far down the waterfall as the role can genuinely influence — opportunity created or pipeline dollars rather than raw lead counts — and audit the stage immediately above any compensated stage for volume inflation.

Small-sample noise. If a stage processes fewer than roughly 30 records a month, month-over-month rate comparison is statistically meaningless. Roll up to quarterly, or track the raw counts and stop computing percentages entirely. Teams that report a 33% rate off three records out of nine make confident decisions on nothing.

The infographic itself becomes the deliverable. The chart gets built, presented once, admired, and never updated. This is the quiet failure mode that ends most implementations. A Demand Waterfall Infographic is a communication layer over a live measurement system; if the measurement system is not instrumented and reviewed on a cadence, you have produced a poster.

A practical rollout plan

Sequence the work so you get a usable signal before you get a perfect model. The wrong order — build the scoring model first, define stages later — produces months of data you have to throw away.

Weeks 1–2: definitions and agreement. Get marketing and sales leadership in one room and write entry and exit criteria for every stage. Argue it out now. Require a named owner per stage and a maximum dwell time per stage. Produce one document, dated and versioned, and have both leaders sign off in writing. Do not touch a single system yet.

Weeks 3–4: instrument the plumbing. Map every stage to a specific field state in your CRM and marketing automation platform, with timestamps written on every transition. Timestamps are non-negotiable — without an entered-stage date you cannot compute velocity, cohorts, or dwell time later, and retrofitting them is nearly impossible. Add a required disqualification reason field with a short controlled picklist rather than free text.

Weeks 5–8: baseline, do not optimize. Let the system run untouched and collect a clean baseline. Resist the urge to change scoring during this window; every mid-stream change invalidates the comparison you are about to need. At the end, compute each stage-to-stage rate, each median days-in-stage, and the same numbers segmented by channel and segment.

Weeks 9–10: build the scoring model against reality. Now that you have outcome data, pull the records that reached SQL and examine what they did before qualifying. Weight behaviors by their actual association with progression. Set the MQL threshold conservatively — deliberately slightly too strict — because a too-loose threshold burns sales trust in weeks and that trust takes quarters to rebuild.

Weeks 11–12: publish the Infographic view and set the cadence. Render the waterfall as a stage-band graphic with volume and rate at each step, and put it on a fixed monthly review agenda with a standing question: which single handoff are we fixing this month? Pick exactly one. Teams that attack four stages simultaneously cannot attribute any improvement to any change.

Ongoing: re-baseline quarterly, audit definitions before investigating anomalies, and run a dead-record sweep every quarter so stalled records get recycled rather than inflating coverage.

The single most common sequencing error is publishing the chart in week two. A waterfall with no baseline invites everyone to declare their own stage healthy, and once that narrative sets, the real numbers arriving in month three read as an attack rather than a finding.

Related questions

What is the difference between an MQL and an SAL?

An MQL is a record that marketing's scoring model says is ready for sales contact. An SAL is a record a human on the sales side has reviewed and explicitly accepted as worth working. The gap between the two counts is the clearest measure of whether sales trusts marketing's qualification.

How many stages should my waterfall have?

Fewer than you want. Five or six stages give you enough resolution to locate a leak without splitting your volume so thin that no stage has statistically meaningful counts. Add a stage only when you have a specific decision you cannot make with the stages you already have.

Does the Demand Waterfall work for account-based motions?

Yes, but you must count accounts rather than individuals throughout, or run two parallel views. Mixing person-level counts at the top with account-level counts at the bottom produces conversion rates that look alarming and mean nothing.

How long before the waterfall shows whether a change worked?

At least one full sales cycle, and usually two before the trend is trustworthy. Changes to top-of-funnel criteria take that long to propagate to bookings, which is why teams that judge a scoring change after three weeks routinely reverse a change that was working.

Should recycled leads re-enter at the top?

Report them as a separate line rather than folding them into new inquiry volume. Folding them in inflates the top of the waterfall and quietly degrades your inquiry-to-MQL rate even when nothing about acquisition actually changed.

FAQ

What exactly does a Demand Waterfall Infographic show?

It renders each demand stage — inquiry, MQL, SAL, SQL, opportunity, closed-won — as a labeled band whose size reflects record volume, with the conversion rate printed on each transition. The visual value is that a non-analyst can find the narrowest handoff in about three seconds, which is exactly what you want in an executive review where nobody will open the underlying report.

How is this different from a standard sales funnel chart?

A sales funnel typically shows only the stages sales owns and treats everything upstream as one undifferentiated pool. The Waterfall makes the marketing-to-sales handoff an explicit, measured stage with its own acceptance criteria and rejection reasons. That handoff is where most organizational friction lives, so making it visible is the whole point of the model.

Do I need a specific tool to build one?

No. The model is a set of stage definitions with timestamps, and any CRM plus marketing automation platform can carry it. The graphic itself can be produced by a BI tool, a spreadsheet chart, or a designed vector asset. Buying a platform before agreeing on stage definitions is the most expensive way to fail at this.

What conversion rate should I be worried about?

Any rate that moves more than roughly double its normal month-to-month variance, or any rate you cannot explain. Absolute thresholds are less useful than deviation from your own baseline, because a "low" rate at one company is a strict-definition artifact and a "healthy" rate at another is a loose-definition artifact.

How often should the waterfall be reviewed?

Monthly for the rates and volumes, quarterly for the definitions themselves. Weekly review is appropriate only for very high-velocity motions where a full cycle completes in days; for anything with a multi-week sales cycle, weekly review means reacting to noise and thrashing the team.

Can this model connect marketing spend to revenue?

It gives you the chain of custody — spend produces inquiries, inquiries produce MQLs, and so on to closed-won revenue — which lets you compute a defensible cost per stage and per closed deal. It does not resolve multi-touch attribution on its own; you still need a stated attribution convention applied consistently across periods.

Sources

• https://www.forrester.com/ — B2B revenue process research, including the demand waterfall model originated by SiriusDecisions
• https://www.gartner.com/en/marketing — B2B buying process and demand generation research
• https://blog.hubspot.com/marketing — lead qualification, lifecycle stages, and funnel conversion practice
• https://business.adobe.com/products/marketo/adobe-marketo.html — demand generation and lead scoring in marketing automation
• https://www.salesforce.com/resources/ — CRM opportunity stages, pipeline management, and forecasting
• https://contentmarketinginstitute.com/ — content strategy and lead nurturing research
• https://www.marketingprofs.com/ — practical marketing metrics and funnel analysis
• https://hbr.org/topic/subject/sales — research on sales process, pipeline management, and buying committees

Related on PULSE

• [The RevOps Funnel — Infographic](/knowledge/gb0009)
• [5 Stages of a Healthy Pipeline — Infographic](/knowledge/gb0010)
• [The Sales Cycle: Lead to Closed-Won — Infographic](/knowledge/gb0011)
• [Demand Gen Review — Title Slide](/knowledge/gb0079)
• [Demand Gen Leader — LinkedIn Banner](/knowledge/gb0031)
• [Demand Generation Lead — LinkedIn Banner](/knowledge/gb0242)