Deciding Which Product to Build Next

A profitable manufacturer with a catalog and a dealer network has a problem a startup does not: too many plausible next products. Sales wants a cheaper version for the accounts they keep losing. The largest customer wants a variant only they will buy. An engineer has a concept in an adjacent category. The owner read about a market growing 18% a year.

All four are defensible. Doing all four is how a company ends up with a bloated catalog, an engineering team that finishes nothing, and a channel that cannot explain the product line. New product strategy is mostly the discipline of choosing, and choosing means writing down what you are not doing.

Allocate the portfolio before you evaluate ideas

Sort projects by how far they sit from what you already do well. Core work improves existing products for existing customers — cost reduction, a new size, a reliability fix — at low risk with modest upside. Adjacent work is one step out: an existing product for a new segment, or a new product for existing customers, where you keep either the market knowledge or the technical knowledge. Transformational means new product, new market, new capability: genuine option value, genuine chance of total loss.

A common allocation is roughly 70% of development spend on core, 20% adjacent, 10% transformational. That is a starting point, not a law — a company in a declining category should shift weight outward.

Writing the allocation down changes the conversation. Ideas stop competing in a single free-for-all and compete within their bucket, so a transformational concept no longer has to beat a safe line extension on payback period. That is the only way transformational projects get funded in a company that is doing fine.

Line extension or new category?

This is the decision most growing businesses face, and the honest answer is usually less exciting than the owner wants.

Two questions cut through it. First: does your existing customer have the problem this product solves? If yes, you have distribution and credibility, and the project is adjacent rather than transformational — cheaper than it looks. If no, you are starting a new business that happens to share a factory.

Second: does your brand have permission? A company known for heavy-duty industrial tools can extend into more of them indefinitely. The same company selling a consumer kitchen product is asking a buyer to accept a claim the brand does not support, and will pay for that in marketing spend. Where a brand can and cannot go is something a proper competitor analysis for a physical product reveals more clearly than internal debate.

A scoring framework you can actually use

Committees argue in adjectives. Scoring forces the argument into numbers and makes explicit which criteria your company really weights. Score each candidate 1–5, then apply a weight:

1. Market size and growth (weight 2). A market that would add 3% to revenue is not worth an engineering program.
2. Strategic fit (weight 3). Does it strengthen the core business or divide attention? Weight this heavily; it is where most bad decisions hide.
3. Channel fit (weight 3). Can your existing sales force and distributors sell it without new training, new relationships, or conflict with what they already carry?
4. Technical confidence (weight 2). Do you know how to build it, or is there real invention risk?
5. Margin potential (weight 3). Realistic gross margin after landed cost, freight, and channel discount — not the optimistic one.
6. Capacity fit (weight 2). Does it use plant, tooling, and suppliers you already have?
7. Defensibility (weight 1). Is there IP, tooling investment, or regulatory clearance that slows a fast follower?
8. Time to first revenue (weight 2).

The output is not a decision but a structured conversation. When two projects score within 10% of each other, the model cannot distinguish them and judgment has to. When one scores 40% below another and someone still advocates for it, ask which criterion the model is missing. Run it retrospectively on your existing products too: if your best seller would have scored poorly, your weights are wrong.

Cannibalization: count it honestly

A new product that takes sales from your existing one has not grown the business by its full revenue. Every portfolio decision needs an explicit cannibalization estimate. Three cases worth separating:

• Deliberate replacement. You are obsoleting your own product on purpose. Fine — plan the transition, the inventory run-down, and the service commitment on the old model.
• Trading down. A cheaper variant captures price-sensitive buyers, but some existing customers switch to it. A product that moves 30% of your base down a tier can cut total profit while raising unit volume.
• Genuine expansion. It reaches buyers who were not buying anything from you. This is what you want, and it is rarer than people assume.

Cannibalizing yourself beats letting a competitor do it. But cannibalizing yourself accidentally, while reporting the new product's revenue as growth, is how a business convinces itself things are going well for two years longer than they are.

Stage gates and real kill criteria

Portfolio discipline fails at the same place in almost every company: nobody kills anything. Projects that should have stopped at concept limp forward because someone senior sponsors them, consuming capacity the good projects needed. The fix is structural — defined gates with defined evidence, and criteria written before the project starts, when nobody is emotionally committed:

What makes gates work is not the meeting — it is writing kill criteria in advance and having someone other than the sponsor apply them. The mechanics are in stage-gate product development with real decision points, and the earliest gate is best served by a proper feasibility study for a product idea rather than an opinion.

One cultural note: a company where nothing is ever killed is also a company where nothing ambitious is ever proposed, because proposing something risky becomes a career liability. Killing a project publicly and treating it as a good decision — money saved, learning captured — is what makes the next transformational idea possible.

Capacity and channel are constraints, not afterthoughts

Capacity. Engineering bandwidth, not capital, is the binding constraint in most growing businesses. Count engineer-months honestly, including the sustaining work nobody schedules but everybody does. A company with three engineers cannot run two development programs and support a catalog.

Channel. A distributor carries a finite number of SKUs and gives you finite attention. A product needing a different sales motion — longer cycle, technical demo, a different buyer inside the customer — often will not get sold by your existing channel no matter what they said in the meeting. Ask them before you tool it. The margin structure the channel needs interacts with volume pricing tiers for B2B customers in ways that can quietly make a new product unattractive to the people who move it.

Test the business model too: a new product may work better as a service, a consumable, or a subscription than a one-time sale, and that changes the development requirements — see choosing a business model for a physical product.

Finally, make it a cadence rather than a document: review the scored list quarterly, put gate reviews on a fixed calendar, and look back annually at what you launched, what it earned, and what you killed. Most companies never check whether their forecasts were accurate, so their weights never get corrected. And when the real answer is not a new product at all, the question in when to build version 2 of your product is often the cheaper win.

Projects House works with established manufacturers on the engineering side of these decisions — feasibility, cost modeling, and turning a chosen concept into a product that fits the plant and the channel you already have. To talk it through, reach us via the contact form.