The NPD process in marketing: eight stages, and what marketing owns at each gate

NPD stands for new product development, and in marketing it describes the eight stage process a product passes through from an idea to a reviewed launch: idea generation, idea screening, concept development and testing, the marketing strategy and business case, product development, market testing, commercialisation, and the post launch review. Each stage is a gate, and marketing's job at every gate is to supply the evidence that answers one question, so the decision to continue or stop is made on something better than enthusiasm. Most products that fail in the market failed at a gate that nobody enforced.

This page covers the process and what marketing owns in each stage. For the plan that carries the launch, see our marketing plan template.

The eight stages, and the gate at each one

Stage	The question the gate answers	Marketing's deliverable	Stop if
1. Idea generation	What problems are worth solving?	A written list of customer problems, sourced from sales calls, service tickets, searches and trade conversations	Nothing on the list comes from a customer
2. Idea screening	Which ideas fit this company?	A screen on market size, fit with the brand, route to market, and capability	You cannot describe who buys it
3. Concept development and testing	Do buyers understand and want it?	A written concept, tested with target buyers, with their words recorded	Buyers cannot say what it replaces
4. Marketing strategy and business case	Does the arithmetic work?	Positioning, price, volume forecast, channel plan, cannibalisation estimate, payback	The margin only works at a volume nobody can justify
5. Product development	Can we actually make it?	Claim substantiation, naming, packaging and label copy, launch content plan	The claim you planned to make cannot be supported
6. Market testing	Does it sell in the real world?	A limited launch with defined success measures: trial, repeat, returns, support load	Trial is fine and repeat is not
7. Commercialisation	Are we ready to scale it?	Launch plan, availability, sales tools, pricing and promotion calendar, search and listing presence	Supply cannot meet the demand you intend to create
8. Post launch review	What did we learn?	Actuals against forecast, reasons for the gap, decisions written down	Nobody books the meeting, which is the usual failure

The three stages teams skip

Concept testing, done honestly

A concept test is not asking people whether they like an idea. It is finding out what they currently do instead, and what they would give up. Run it this way:

• Write the concept in one paragraph: who it is for, the problem, what it does, and the form it takes. No adjectives.
• Show it to eight to twelve people in the target group, one at a time.
• Ask what they do about this problem today, and what that costs them in money, time or risk.
• Ask them to explain the concept back to you. If they cannot, the problem is the proposition, not the audience.
• Ask about price last, and ask it as two questions: what would you expect this to cost, and at what price would you stop considering it.
• Count the people who ask how to buy it. That number is worth more than every rating scale in the test.

The business case

Five numbers, each with the assumption behind it written down: price, unit cost, forecast volume by period, the cost to launch, and the effect on existing products. The cannibalisation line is the one most often left out and the one that most often turns a good launch into a flat quarter. State it as a range and say which existing product loses the volume.

The post launch review

Book it on the day you approve the launch, for ninety days after it. Bring the forecast and the actuals, name the biggest gap, and write down one decision. Teams that do this get better at forecasting within three products. Teams that do not repeat the same optimism for years.

Where new products are genuinely hard to market

A new product in an existing category has demand waiting for it. A genuinely new product does not, and the mistake is to market it with the name you invented. Nobody searches for a category that does not exist yet.

• Write the pages against the problem phrasing, which people do search for, and introduce your category name inside them.
• Describe the product by what it replaces: the job it does, in the words customers already use for that job.
• Expect to buy or earn the first wave of attention rather than harvest it, and budget for that period explicitly.
• Claim the category name everywhere at the same time: your site, your listings, your profiles and your documentation, so the name becomes consistent early.

The launch artefacts worth having before day one

• A positioning statement naming the target buyer, the frame of reference, the point of difference and the proof. One paragraph, agreed by sales and product.
• Substantiated claims. Every performance or comparison claim with the evidence filed beside it, and the wording legal has approved.
• Pricing and promotion plan, including what the introductory offer is and when it ends, decided before launch rather than during.
• Channel and listing plan. Who sells it, what each one needs, and the data feed or product content each channel requires.
• Sales tools. One page for the buyer, one objection handling sheet for the seller, one demonstration script.
• The content set, written to the problem, published before the launch date so it has time to be found.

The numbers that tell you a launch is working

• Trial. How many people bought once. This is a function of awareness and availability.
• Repeat. How many bought again inside the expected cycle. This is a function of the product.
• Weighted distribution. Not how many outlets carry it, but how much of the market those outlets represent.
• Awareness within the target group, measured the same way each time.
• Returns and support contacts per hundred units, which surface a product problem before the repeat number does.
• Gross margin after launch costs, read against the business case, not against enthusiasm.

Trial and repeat together give you the diagnosis. Low trial with good repeat is a marketing and availability problem. Good trial with low repeat is a product problem, and no amount of advertising fixes it. The positioning work that underpins all of this is in segmentation, targeting and positioning, and the demand side of what you are building for is in needs, wants and demands.

Governance: who signs, and how to kill a product politely

A stage gate process only works if somebody can say no. Name one decision group, give it the four standing questions below, and record the decision in writing after every gate.

1. What evidence do we have that was not here last time?
2. What has changed in the business case?
3. What would have to be true for this to fail, and is it true?
4. Continue, change, hold or stop?

Make stopping normal. A pipeline where nothing is ever stopped is not a pipeline, it is a queue, and the cost shows up as launches nobody believed in.

Questions buyers ask about NPD process in marketing

What does NPD stand for?

New product development. In marketing it refers to the structured process from idea to launch and review, usually taught as eight stages with a decision gate at each one.

What are the eight stages of the NPD process?

Idea generation, idea screening, concept development and testing, marketing strategy and business case, product development, market testing, commercialisation, and post launch review. Some textbooks merge the fourth into the third or split development into design and engineering, and the sequence is the same.

What does marketing own in the NPD process?

The evidence at every gate and the launch itself. Specifically: the customer problem list, the screening criteria, the concept test, the positioning, the price and volume forecast, claim substantiation, the launch plan, and the post launch review.

How long should an NPD process take?

Set the calendar from the gates rather than from a target launch date. A software feature can pass all eight in weeks; a regulated physical product takes much longer because testing and approval sit inside stage five. The useful discipline is a date for each gate, not a date for the launch.

Can a small business run a full NPD process?

Yes, at a smaller scale. Twelve conversations instead of a research programme, a one page business case instead of a model, and a single market test with one retailer or one customer segment. The gates are the valuable part, and they cost nothing but the discipline to hold them.

What is the difference between market testing and a soft launch?

Market testing is a limited release designed to produce a decision, with success measures agreed beforehand. A soft launch is a full launch with the publicity turned down. If nobody wrote down what would make you stop, you are doing the second one.

At the launch gate, ask the assistant who already owns the category

By the commercialisation gate, marketing has tested concept, price and positioning with real buyers. One test remains cheap: ask an assistant what it recommends for the need your product serves.

The reply shows the products a buyer will compare against yours on day one, and the language it uses to describe them. AI Knows Us runs your category question across the main assistants, so the launch plan starts from what they actually say.