From funnel to forecast: sizing the gaps

A funnel is a forecast in disguise

Once you know your step rates and your average order valueAverage order valueThe average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".View in glossary, the funnel stops being a diagram and becomes arithmetic:

traffic × rate₁ × rate₂ × … × average order valueAverage order valueThe average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".View in glossary = revenue

Change any input and the model tells you, in pounds, what the change is worth. That single property turns funnel analysis from a reporting exercise into a planning tool, because every idea competing for next quarter's budget can now be priced with the same formula.

Price the levers, with the maths shown

Take a worked baseline month for an online store:

• 50,000 visits, 4% add to basket: 2,000 baskets
• 60% of baskets reach checkout: 1,200 checkouts
• 70% of checkouts complete: 840 orders
• 840 orders × £45 average order valueAverage order valueThe average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".View in glossary = £37,800 a month

Now price one realistic move on each lever, holding the others still:

• Traffic +10% (5,000 extra visits): everything downstream scales, so +84 orders = +£3,780
• Basket rate 4% → 5%: 2,500 baskets, 1,500 checkouts, 1,050 orders. +210 orders = +£9,450
• Checkout completion 70% → 75%: 840 → 900 orders. +60 orders = +£2,700
• Average order valueAverage order valueThe average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".View in glossary +£5: 840 × £5 = +£4,200

The ranking is rarely what the meeting expected. One percentage point on the basket step is worth two and a half times a 10% traffic push. Put another way: matching that single point through traffic alone would need 12,500 extra visits, a 25% increase, bought again every single month. The conversion gain, once won, keeps paying at no extra cost.

Two cautions before you fall in love with a lever. First, price realistic movement: a point on a 4% step is plausible, a point on a 95% step barely exists. Second, levers interact: a discount that lifts conversion can lower average order valueAverage order valueThe average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".View in glossary, so model the pair together, never in isolation.

From priced levers to a forecast

A forecast is the same model run forwards with the moves you actually plan to make. Baseline £37,800; the quarter's plan is the checkout fix (+£2,700) and the basket-page work (say half a point, +£4,725). Forecast: around £45,000 a month by quarter end.

Except never say "£45,000". Your rates wobble month to month, so build the model on your last three months' range, not your best month, and forecast a range: at the low end of recent rates the same plan lands near £42,000, at the high end near £48,000. A range signals you know the difference between a model and a promise; a suspiciously precise number signals the opposite.

Defend it in one line

The end product of all this is a sentence. Something like: "At today's rates, 12,500 extra visits and one point on the basket step are worth the same £9,450, and the second one we only have to win once." That is a funnel forecast doing its real job: making the trade-offs visible in pounds, on your own numbers, so the argument is about evidence instead of enthusiasm.

It also changes the conversation when things go wrong. If the quarter lands below forecast, the model shows which input missed: traffic arrived but the basket rate never moved, or the rate moved and traffic fell away. Instead of "marketing missed its number", you get "the checkout fix worked, the basket test did not, here is what we learned". A funnel you can forecast with is a funnel you can be honest with.