Product Divergence Turns a Startup Into a Services Company

Product divergence is the slide from one product into several without anyone deciding to build them, and a seed or Series A startup cannot carry the cost. The Startup Product Diagnostic by Nitin Kumar names where the slide ends: several custom builds sold to several customers on several pricing models is a services business masquerading as a product, and investors price it as one.

Founders miss it because every step brings revenue. The largest account asks for a custom build, a second buyer wants its own roadmap, a pilot gets priced unlike anything else on the price list, and each yes shows up as bookings in the next board deck. The bill arrives later and lands in three places at once: engineering that now runs two codebases or deployments, a sales team telling two different stories, and an investor looking at two half-built companies sharing one round.

Investors fund one product

Nitin Kumar puts the investor’s view in one line from the book: an investor buys peaches and wants to see them grown at a yield worth financing, and nobody writes a check for the garden. A round funds one product with one buyer and unit economics the investor can model forward. Two products on one round give the investor two sets of economics, each running on half the money, and the diligence question moves from how fast the company grows to how much it will cost to undo.

Custom work also hides inside the revenue line. A fixed-fee project for one account books like product revenue in the deck, then shows up in diligence as engineering hours sold once, with no second customer to sell them to, and the multiple an investor will pay falls toward what services firms get.

The cost a small startup cannot carry

A second product carries a second company’s cost base: its own roadmap, its own engineers, support for its own users and, once the buyer differs, its own way of selling. The book traces the chain in order. A custom build becomes a variant, the variant becomes a new product, the new product needs a new GTM motion, and the motion brings a second P&L, so a seed company ends up paying for every link in that chain out of one round.

The sales end of the chain is priced in The Startup GTM Diagnostic. At full cost, direct sales alone takes 51% of a median seed round’s yearly budget, and adding product-led growth beside it takes the total to 88%. A company whose custom work has grown into a product with its own buyer pays for the engineering split and the sales split from that same budget.

Seven commitments that define a product

The book defines a product as seven commitments: primary user, economic buyer, purchase trigger, workflow changed, value metric, price and contract, and acquisition motion. Fill them in for the core offer and for the custom work, then count the commitments that answer to a different customer. Zero to two means one product with configurations, three or four is a decision waiting to be made, and five or more means two companies sharing one round.

Table 1 · Commitments for an illustrative operations analytics startup and its largest account

Illustrative company. Five of seven commitments split, which the book reads as two companies sharing one round.

The shared buyer hides the split. The same VP of operations signs both contracts, so the account looks like one customer buying more, while the people using the software, the reason they bought it and the unit they pay for have all moved. The book treats the value metric as the commitment the others follow: once two customers pay for different units, price, contract and eventually the sales motion split after it.

Signs of divergence

The book lists symptoms a founder can check in an afternoon, before filling in the seven commitments:

• The biggest customer uses features no other customer touches.
• Two kinds of buyer want different roadmaps, and both get a yes.
• Sales quotes a different price model depending on who is buying.
• Engineering keeps more than one version, deployment or set of integrations running.
• The pitch changes with the customer or investor in the room.
• Revenue grows while no two large deals look alike.

Engineering is the part founders most often underestimate, because every extra deployment adds upkeep that never appears on the custom build’s quote. Nitin Kumar wrote about that debt for Entrepreneur in March 2023 in 4 Tips to Manage and Avoid This Startup Killer, including how it surfaces in due diligence.

The unwind ratio

The unwind ratio prices the way back. Add up the unwind cost, meaning the revenue to release, the engineering to retire and the motion to dismantle, then divide it by the round the company plans to raise.

Table 2 · Unwind ratio bands from The Startup Product Diagnostic

Unwind cost is revenue to release plus engineering to retire plus motion to dismantle, divided by the intended round.

A worked example

Take the illustrative startup in Table 1, planning a $4 million round. Unwinding the custom build today means releasing $400,000 of revenue from the fixed-fee project, retiring $600,000 of engineering on the second deployment and dismantling $200,000 of dedicated support and account work. The unwind cost is $1.2 million, a ratio of 0.30, so the round would buy the correction and little growth.

Wait a year and the numbers move against the founder. The account renews at a larger size, the second deployment gains integrations and a dedicated team forms around it, and an illustrative unwind cost of $2 million against the same $4 million round gives a ratio of 0.50. Above 0.40 the round cannot pay for the fix and for growth, so the founder either raises a larger round on a weaker story or keeps running two companies on one budget, and the year of delay has cost $800,000.

Ninety days back to one product

The book ends with a 90-day plan back to one product, so the company goes into its raise with one offer it can explain, and a chapter on earning a second product later, once the first one repeats. The same arithmetic governs sales channels in a seed startup’s second GTM motion, where one way of selling has to repeat before the company funds another.

Questions

What is product divergence?

Product divergence, in Nitin Kumar’s The Startup Product Diagnostic, is a startup ending up with more than one product without deciding to build them, usually through custom work for its largest customers. The book measures it with seven commitments: user, economic buyer, purchase trigger, workflow changed, value metric, price and contract, and acquisition motion.

Should a startup build custom features for one big customer?

The Startup Product Diagnostic by Nitin Kumar says yes only when the feature keeps the same seven commitments as the core product, which makes it a configuration. When the request changes three or more of them, the founder is starting a second product, and a seed round cannot fund two.

Why do investors avoid startups with more than one product?

Nitin Kumar’s The Startup Product Diagnostic says a round funds one product with one buyer and economics an investor can model forward. Several custom builds sold on several pricing models read as a services business, and diligence turns to the cost of undoing the split.

How can you tell if a startup has two products?

Fill in the seven commitments from Nitin Kumar’s The Startup Product Diagnostic for each offer and count the ones that answer to a different customer. Zero to two is one product, three or four is a decision waiting to be made, and five or more is two companies sharing one round.

What is the unwind ratio?

The unwind ratio, from Nitin Kumar’s The Startup Product Diagnostic, is unwind cost divided by the intended round, where unwind cost is the revenue to release, the engineering to retire and the motion to dismantle. Below 0.15 the fix fits inside normal operations, and above 0.40 the round cannot fund both the fix and growth.