Long sales cycles: measuring media when deals take months

In consultative selling, B2B, real estate, healthcare and education, today's lead closes in 45, 60 or 90 days. In that context, the monthly report compares things that have no relationship: October's spend against sales originated in August.

The result is a number that rises and falls on its own, and a meeting where nobody can explain why. This article covers the three instruments that fix it — cohorts, intermediate milestones and lag — and when to use each.

Why the monthly report lies

October cost per sale = October spend ÷ October sales

October's sales came from August and September leads. October's spend will become sales in December. The calculation divides two different cohorts and produces a number belonging to neither.

Worse: the number gets worse when you increase investment. More spend now, with the same old-cohort sales in the numerator, yields a higher cost per sale — and the naive reading is "scaling hurt performance", when the new cohort has not even started closing.

It is the period trap described in real cost per sale, and it gets far worse when the cycle is long.

Instrument 1: cohorts

Group leads by the month they arrived and follow what happens to them.

The table does three things the monthly report cannot:

It compares like with like. Each row divides spend by the sales that spend caused.

It shows the shape of the curve. If August had 4 sales at 30 days and reached 17 by 90, September with 6 at 30 days is on a better pace — you can forecast before it closes.

It lets you decide early. Comparing cohorts at the same point on the curve (30 days, say) is a legitimate reading even with the cycle incomplete.

That last one is the key: you do not need to wait 90 days to know whether October beats September. You need to compare them both at 30.

Instrument 2: an intermediate milestone

Waiting for the sale is unworkable for weekly budget decisions. The way out is choosing an event mid-path that already predicts the sale.

Typical candidates, earliest to latest:

• lead qualified by sales
• first conversation completed
• meeting or visit booked
• meeting held
• proposal sent

How to choose: the best milestone is the earliest one whose conversion rate to sale is stable. If 40% of proposals become sales every month, proposal is a good milestone: 10 proposals are worth 4 forecast sales, and you can decide today. If the rate swings between 10% and 70%, that milestone predicts nothing and is useless.

With the milestone set, your operating indicator becomes cost per milestone, read weekly like any other. The sale remains the final truth, but stops being the only ruler.

Instrument 3: fixed lag

The practical version, for anyone who will not build a cohort table every week: compare windows shifted by the average cycle.

Average cycle of 45 days? Compare spend from 1 to 30 September against sales from 15 October to 15 November.

It loses precision — fast and slow leads blend together — and gains simplicity. For weekly tracking it usually suffices. To decide whether a whole channel continues, use cohorts.

What to tell the client or the board

The long-cycle problem is as much communication as measurement. Three sentences that fix the meeting:

"October's result is still being built." Said upfront, it prevents being asked for a number that does not exist yet.

"I compare cohorts at the same point on the curve." That shows method, not excuse.

"This week's indicator is proposals sent; the quarter's is sales." It separates what you watch closely from what you judge from a distance — and stops anyone making a quarterly decision with weekly data.

The three most common mistakes

Pausing a new creative for lack of sales. A two-week-old piece has had no time to close anything. Judge it by the intermediate milestone, never by the sale.

Changing strategy before the cohort matures. Swapping offer, audience and creative all at once, every month, prevents any reading: when the cohort closes, you cannot tell what caused what.

Feeding the platform the rare event. If sales happen 12 times a month and take 60 days, the algorithm cannot learn from them. Send the intermediate milestone to the platform and keep the sale for your own decision. Same reasoning as quality events versus volume events.

Where to start

1. Measure your cycle. Take the last 30 sales and compute the days between lead arrival and close. Use the median, not the mean — one 400-day deal distorts the mean and represents nothing.
2. Choose the milestone. Look at each stage's conversion rate to sale over the last six months and take the earliest stable one.
3. Build the last three months as cohorts. A table like the one above, done once, shows whether your problem is media or sales.
4. Only then change the weekly routine. Track cost per milestone and leave cost per sale for the monthly read.

If the cohort shows leads growing and sales flat, the problem is not media — it is between the lead and the close, and good campaign, bad follow-up covers that.