Build a Marketing Funnel Backwards From a Revenue Target

Build a funnel forwards from a budget and you find out whether the target was reachable some time in month four. Build it backwards from the revenue target and you find out in ten minutes — because the arithmetic tells you how much traffic the plan requires, and that number is usually the thing nobody had looked at.

The method is one line: divide the target by every conversion rate between the money and the visit. Everything else is knowing which rates to use and being honest about them.

The arithmetic

Working from revenue back to sessions:

orders needed      = revenue target ÷ average order value
sessions needed    = orders needed ÷ conversion rate

For a lead-generation business there are more stages, so more divisions:

sales needed       = revenue target ÷ average deal value
opportunities      = sales needed ÷ close rate
raw enquiries      = opportunities ÷ qualification rate
sessions needed    = raw enquiries ÷ enquiry conversion rate

Each division makes the number bigger, which is why the answer so often lands somewhere nobody expected.

Worked example 1: ecommerce, $50,000 a month

Assume an average order value of $120, a site-wide conversion rate of 1.8%, a gross margin of 40%, and that 60% of sessions will have to be bought at $1.20 a click.

That gives a ROAS of $50,000 ÷ $16,680 = 3.0×. Now check it against margin: gross profit is $50,000 × 0.40 = $20,000 against $16,680 of media, leaving $3,320 before any salary, fee or platform cost.

The plan technically hits the revenue target and does not survive contact with the rest of the cost base. That is a finding worth having in week one, and the forwards version of this calculation would not have produced it. A 3.0× ROAS would have been reported as a good result all year, which is why ROAS needs a margin figure beside it before it means anything.

Worked example 2: lead generation, $600,000 a year

Assume an average deal of $10,000, a close rate of 20%, a qualification rate of 40% — four in ten enquiries are worth a sales conversation — and an enquiry conversion rate of 3% on the site.

$600,000 a year is $50,000 a month, so:

Five sales a month requires roughly 2,083 sessions to the pages that convert — not 2,083 sessions to the site. That distinction is where most of these plans quietly go wrong, because blog traffic and pricing-page traffic convert at rates that differ by an order of magnitude.

If the site currently gets 700 such sessions, the gap is 1,383. Buying it at $4 a click costs $5,532 a month and produces, at the same rates, 1,383 × 0.03 × 0.40 × 0.20 = 3.3 sales — 3.3 × $10,000 = $33,200 of revenue for $5,532 of media. That is a ROAS of 6.0×, and now the question is whether the sales team can handle 63 enquiries a month, which is a different constraint entirely.

The rate that costs nothing to improve

Lift qualification from 40% to 55% — better form questions, clearer pricing, a disqualifying line in the ad copy — and the arithmetic changes:

25 ÷ 0.55 = 45.5 raw enquiries, and 45.5 ÷ 0.03 = 1,517 sessions.

27% less traffic for the same five sales, from a change with no media cost. This is the general result: the further up the funnel you fix something, the more traffic it saves you buying, and the rates nearest the money are usually the cheapest to move.

A qualification rate is a poor thing to make a target, though. It can also be lifted by relabelling enquiries that were never going to buy, which produces the same percentage and none of the sales — so in a report it belongs as a guardrail rather than a headline, paired with the enquiry-to-meeting rate that would fall if the definition had quietly loosened.

Which rates to use

The honest answer is your own, from the last 90 days, and nothing else. Three cautions:

• Do not use published benchmarks. A conversion rate averaged across an industry is a number from businesses with different traffic, different prices and different buyers. It will be wrong for you in a direction you cannot predict.
• Use a period long enough to contain a full sales cycle. A 90-day window on a business with a four-month cycle is measuring the wrong months against each other.
• If you genuinely have no data, run the model three times — pessimistic, expected, optimistic — and look at whether the target survives the pessimistic case. That is more useful than one guess presented as a forecast.

When the answer is “this target is not reachable”

Sometimes the required traffic is a multiple of anything the market can supply. That is the most valuable output this exercise produces, and there are only four responses:

1. Change the target. The most honest and least popular.
2. Change the price. A higher average order value divides the same revenue into fewer orders, and fewer orders need less traffic.
3. Change a conversion rate. Cheapest per unit of effect, and slowest to move.
4. Change the channel mix. Only helps if the new channel has a genuinely different cost per session, not just a different name.

Notice that “spend more” is not on the list unless the reach exists to spend it on. Reach, not budget, is the binding constraint more often than people expect — which is the whole reason to do this backwards.

Turn it into a budget

Once the traffic number exists, the budget follows from it rather than the other way round: sessions needed by channel, multiplied by cost per session, plus the fixed costs. The funnel calculator does this across all six stages, for four different funnel shapes, and shows which stage is the binding constraint — then the budget template turns it into a twelve-month budget by channel, with a forecast tab that works the same direction as this post and ends in a viability check.

If the calculation says a conversion rate has to move, start with the CRO checklist, which scores ideas by ICE and flags any test with too few conversions per arm to trust the result.