Product-Led Growth: The Arithmetic That Decides It

What product led growth promises

Product-led growth means the product does the selling, and that definition is not in dispute. What no definition prices is the thing being replaced, and that thing has a published price.

The product drives acquisition, retention and scale, sometimes with activation and expansion added. All of those versions claim the product does the primary work a salesperson would.

So the definition is settled. The decision is not, and it sits further down the B2B marketing funnel than the definitions suggest.

Check who is doing the defining before you weigh the agreement. A definition attached to a framework somebody sells is not independent evidence that the framework fits your business.

A definition that never names a failure case is a description, not a test. Treating product-led and sales-led as opposites is the other common framing, and this page argues with it.

Across the definitions, the shared claim is that the product can remove a salesperson from the purchase. A salesperson has a market price, and it is published every two years.

The Bridge Group, read 7 September 2026: what one salesperson costs and carries

1. 1Median account executive on-target earnings reached $200,000, up from $190,000 in 2024.
2. 2Median annual quota is $960,000, which is 4.8 times the median on-target earnings figure beside it.
3. 3Quota attainment fell to 48 percent in 2026, down from 51 in 2024 and from a band around 66 percent between 2015 and 2022.

Two numbers matter here and the second one is the lever.

On-target earnings tell you roughly what a rep is paid, and the employer cost on top is not published here. Quota tells you what a rep has to bring in for the arrangement to hold.

Divide that quota by your annual price and you get a customer count. That count is the whole decision.

Work out what one salesperson must close

One account executive is expected to bring in $960,000 a year. At a high price that is a handful of deals. At a low price it is a much larger number, and seeing how much larger is the first useful thing this arithmetic does.

Customers per rep

The sum has one step. Take the median quota and divide it by what one customer pays you in a year.

Divide each of those by 250 working days. Sixteen deals a year is one every fifteen working days. Sixteen hundred is between six and seven every working day, closed one at a time, all year.

The same division runs the other way for anyone comparing this against B2B lead generation cost per lead, where the unit is a lead and the salary is still in the sum.

No published research prices the ceiling on deals one person can close in a year, and none is claimed here. Your own sales records hold that number, and the daily rate above is what you hold it against.

This reframes the usual question of whether you should choose product-led growth at all. Dividing first tells you what your price leaves available to choose from.

Signups behind the customers

Removing the rep does not remove the cost. It moves the cost to the top of the funnel, and the exchange rate there is published too.

OpenView, 7 June 2022: five percent of freemium signups pay

1. 1Only 5 percent of freemium signups convert to paid customers.
2. 2Median visit to signup rate is 6 percent, so the traffic behind the signups is another multiple again.
3. 3Companies that contacted their users retained them 2 to 3 times better, which is the first crack in the pure version of the model.

Now take the customer counts from the chart above and divide each one by that rate.

A pricing page with both doors on itWhat the model looks like to the person deciding

The difference between the two models is what the page asks for. Two fields and nobody in the path, or seven fields and a person who calls you back. Whichever door the money walks through is the model you are running, whatever the deck says.

Run it on your own price

Four inputs, and three of them are already in your billing system. Run this before you rebuild anything, because the answer often removes the question.

Six steps, in order.

1. Write down your annual price

   What one customer pays you across twelve months, at the plan most of them buy. Not the enterprise tier on the pricing page.

2. Divide 960,000 by it

   That is how many customers one salesperson would need to close to carry a median quota. Divide it again by 250 working days to see the daily rate the job implies.

3. Find your own conversion rate

   Divide the paying customers who came from last year's signups by the total number of those signups. With no free tier there is no rate yet, and the published 5 percent measures freemium products only, so borrowing it would answer a different question.

4. Divide the customer count by that rate

   You now have the annual signups the product would need to do the work of one rep. Compare it against what you got last year.

5. Price the traffic

   Divide the signups by your visit-to-signup rate to get visits, then price those visits at whatever acquisition costs you today. Compare the shape against the SaaS marketing funnel if you want the stage names. That is the real bill.

6. Check what the visits land on

   The signup rate in step three is a property of the page as much as the traffic. A SaaS landing page has its own published median, so a low rate may be the page and not the market.

7. Compare the two bills

   One salesperson at $200,000 of on-target earnings, before whatever employer costs sit on top, against the traffic bill from step five. Neither figure is published for your business, so this is the step where your own numbers decide it.

Read this for shape, not for a verdict.

At a low price the daily deal rate is the first thing to weigh, before any cost comparison starts. At a high price the deal count stays small and the comparison becomes a straight cost question.

Nothing published here says where the crossover sits, because the missing input is yours: what one signup costs you to buy. This division narrows the question to that one number, and then hands it back.

How often free users pay

The 5 percent figure is used as though it were a constant. That figure is one measurement of one model in one year, and the published spread around it changes the arithmetic by a factor of ten.

Rows three and four are two bands from the same collection. The six rows overall cover six populations. That is how you can tell the six rows are six populations and not one series.

Read the bottom two rows against the top one. The reported 30 percent rate is six times the reported 5 percent rate, and the two studies measured different populations. Asking for a card is a product decision, made before any of this.

The mirror image of this sum is what account-based marketing costs, where the list is short, the price is high, and the person stays.

So the rate in step three of the checklist is not a benchmark you meet. It is a consequence of a choice you already made about how people start.

Where the human comes back

The purest reading of product-led growth removes people from the funnel entirely. The benchmark data disagrees with the purest reading.

Quadrupling a 5 percent rate takes 32,000 signups down to 8,000. Compare that against the cost of the people making the contact, and the sum from the checklist runs again with a different answer.

Suppose the signup half is cheaper for you to move: that is our SaaS SEO service.

Where product led growth breaks

Four failure conditions decide whether the model fits. A fifth comes out of the model above, and it is the one a founder is most likely to get wrong, because it looks like a preference and behaves like a wall.

The first row deserves a number and the published one is weak. One threshold in circulation is ten minutes to first value, and sixty seconds for products launched into an agent-driven market.

No study sits behind either figure. Treat both as unsupported practitioner heuristics, and measure your own instead.

Client-work failure lists add qualitative conditions and no thresholds. These qualitative lists give you no threshold to hold your own product against.

A sixth failure has no threshold at all and shows up in every set of examples. Slack, Dropbox and Calendly are cited on most of these pages as proof the model works. All three had free distribution at a scale a new company cannot buy, so they explain the outcome and not the method.

What to watch after launch

Three numbers, and the third one is where most reporting goes wrong. The first two are already on every product dashboard in the category. The third has to be worked out once a year, and only that one answers whether a person belongs in your sale.

Report the signup count beside the rate, every time. A 4 percent conversion on 200 signups and a 4 percent conversion on 20,000 are different facts wearing the same number.

If you split signups by behaviour before counting them, the unit you are counting is what a product qualified lead is, and the rate above will move for that reason alone.

Print the denominator.

Then rerun the division once a year. Your price changes, your conversion rate changes, and the answer to whether a person belongs in your sale changes with them.

Everything above the signup is a separate bill, and what B2B demand generation rests on is where that side gets measured.

The traffic side of that bill has its own arithmetic: how much a published article has to average before it replaces a salesperson is the subject of SaaS content marketing.