पाठशाला Pathshala · उत्पाद Utpād, The product · Lesson 24 · Scale

The second product: when and how to make the bet

A second product is a new company inside the old one. Judge it by how close the first is to saturation, whether it sells to the same customer, and whether the team can carry both.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

The idea for a second product usually arrives at the worst moment: growth in the first has slowed for a quarter, a large customer asks for something adjacent, and a new product looks easier than the hard work on the old one. Sometimes it is the right call. More often it splits a team that was not finished with its first job.

This lesson gives three tests for when a second product makes sense, the arithmetic of what it costs the first, a worked example, and a way to run the decision as a bet with a budget and a kill criterion rather than as a strategy that cannot be undone.

Why the second product tempts too early

Three pressures push founders towards a second product before the first is done. Slowing growth looks like a ceiling when it is often a problem in activation, retention or a channel that has run out. Customer requests for adjacent products feel like demand when they come from the few customers who would buy anything the founder sold. Investor stories about platforms and expansion revenue make a single product sound small. None of these is evidence that the first product has reached its limit.

The cost is easy to underestimate because it does not appear on a budget line. A second product needs its own discovery, its own [path to product-market fit](/library/what-product-market-fit-looks-like-in-the-numbers), its own onboarding and support, and the founder’s attention, which is the scarcest resource the company has. It also starts with the same odds every new idea has. Ronny Kohavi and colleagues reported that at Microsoft only about one third of ideas improved the metrics they were built to improve. A second product is a large idea, not an exception to that rate.

Test one: is the first product near saturation?

Saturation is measurable. Take the number of accounts the first product can realistically reach, the bottom-up count from [your market sizing](/library/market-sizing-an-investor-will-believe), not the slide version, and divide paying customers by it. Then compute how many months it would take, at today’s rate of net new customers, to reach half of that market. If penetration is under a third and the headroom is more than three years, the first product has room, and slowing growth is a problem to diagnose, not a signal to diversify.

Saturation can also be qualitative. The product may have won most of the customers who match the [ideal profile](/library/ideal-customer-profile-on-one-page) and the next segment needs a different product anyway. Revenue per customer may have stopped growing because customers use everything the product does. These are real signals. A quarter of slow sales is not one.

Test two: the shared customer

The best second products sell to a customer the company already has, through a channel it already runs. The trust, the billing relationship, the sales team’s calendar and often the data are already paid for, so the second product starts with an advantage no new company would have. The question to answer is whether the same person buys, or at least the same organisation through the same door.

Freshworks shows the pattern. Its two flagship products are Freshdesk for customer service and Freshservice for IT service management, and the company says it is trusted by more than 74,000 businesses. One serves the support team and the other the IT team: a different buyer, often inside the same kind of company, with the same underlying idea of a ticket that has to be resolved. Zoho began as AdventNet in 1996, says it has never taken money from investors, and prefers to build its products in-house rather than grow through acquisitions. The common thread is a customer the company already knows, not a market it admires.

Test three: can the team carry both?

A second product needs a team that wakes up thinking only about it. Amazon’s two-pizza teams give the minimum: a group of fewer than ten with single-threaded ownership of one product across its whole life. Borrowing engineers from the first product for a few days a week fails, because the first product’s urgent work always wins. So the test is concrete: can the company move five to eight people, including someone who can act as the product’s founder, without the first product’s roadmap falling below what its customers need?

If the answer requires hiring the whole team first, the bet is larger than it looks, because the company is also betting that it can hire and integrate a team while running the first product. That is sometimes worth it. It should be priced as such.

At the defaults, 2,400 customers in a reachable market of 12,000 adding 60 a month, the first product has 20 per cent penetration and five years of headroom to half the market. Moving a quarter of the engineers to a second product bought by a fifth of existing customers at 40 per cent of the price produces about ₹29 lakh a month by month 24, and the company ends ahead by a little against staying focused. Lower the attach rate to 10 per cent and the second product loses. Raise net new customers to 120 and the cost of slowing the first doubles. The figure rarely gives a large win in two years; the case for a second product is about what happens when the first one stops growing, which is why test one comes first.

A second product borrows the first one’s customers, channel and team. Make sure the first can spare all three before you take them.

A worked example: school software in Ahmedabad

An Ahmedabad company sells fee collection and accounts software to private schools in Gujarat and Rajasthan. It has 1,100 schools paying an average of ₹12,000 a month, adds about 25 a month, and estimates 4,500 schools in its two states that are large enough to buy. Penetration is 24 per cent; headroom to half the market at today’s pace is nearly four years. Several schools ask for a bus-tracking product for parents.

Test one fails: there is room in the first product and new-school sales have slowed because the company has not yet entered Maharashtra, which is a sales problem. Test two passes: the buyer is the same school administrator and the parents are already on the fee app. Test three is marginal: the team of 22 engineers could move six, but the fee product’s GST and audit features are already behind. The decision is to wait two quarters, enter Maharashtra, and in the meantime run a [concierge test](/library/concierge-and-wizard-of-oz-mvps) of bus tracking with ten schools using an off-the-shelf tracker and a manual WhatsApp update to parents. If seven of the ten pay for it at the end of the quarter, the bet goes to the board with its own team and budget.

Making the bet

When the tests pass, run the second product as a bet in the sense of the [roadmap lesson](/library/roadmap-as-set-of-bets). Write the hypothesis: which existing customers will buy, at what price, how many in the first three quarters. Give it a separate team with a single owner, its own metric and a fixed budget in engineer-months. Write the kill criterion before the first line of code: if fewer than a stated number of customers pay by the end of the third quarter, the team returns to the first product and the second is closed or sold. Sell it first to existing customers who did not ask for it, because their behaviour is better evidence than the requests of those who did.

Two smaller decisions follow. Price it on its own first. A bundle hides whether anyone values the second product, so sell it separately for the first three quarters and offer the bundle only once the standalone number is known. Tell the first product’s customers what will not change. They will notice that some engineers have moved; a short note on what the first product’s roadmap still holds keeps the base that the second product depends on.

The half-yearly portfolio review

Twice a year, before the annual plan and at the half, recompute three numbers for the first product: penetration of the reachable market, months of headroom at the current pace, and the trend in revenue per customer. For any second product in flight, read its metric against the hypothesis and apply the kill criterion without negotiation. Then ask the three tests again of any new idea on the list. Write the answer in a page the board can read, with the numbers, and keep it beside the last one. A company that makes the decision this way will start fewer second products and finish more of them.

The figure is a simplified model and the Ahmedabad company is illustrative. Company figures are as stated on their sites, checked 10 October 2026.

Sources

1. Ron Kohavi et al., Online Experimentation at Microsoft, 2009 — Only about one third of ideas improve the metrics they were designed to improve.
2. Freshworks, About us (checked 10 October 2026) — Freshdesk and Freshservice as flagship products; 74,000+ businesses.
3. Zoho, About us (checked 10 October 2026) — Began as AdventNet in 1996; has never taken money from investors; builds products in-house.
4. Daniel Slater, Amazon’s two-pizza teams, AWS Executive Insights