Why Corporate Innovation Stalls at Launch

Corporate innovation teams fail at commercialisation, not at invention. Blazon has watched this happen inside large companies for a decade, and the work almost never arrives as a technical problem.

The prototype exists. The testing is done.

What is missing is a launch. Launch capability lives outside the innovation function: brand, PR, paid media, commerce and retail sit inside business units whose targets are this year's numbers on existing lines. A new product from the innovation team arrives there with no P&L, no shelf space, no media budget and nobody whose bonus depends on it.

It does not get rejected. It gets deprioritised, which is slower, quieter and much harder to argue with.

That is not a theory. A vice president at Mondelēz International said it on the record in 2025, and the closure notices filed by innovation programmes run by some of the largest companies in the world over the past decade say the same thing in a different register.

The gap is between "it works" and "it sells"

An innovation unit is usually staffed for discovery. It has researchers, engineers, designers, a prototyping budget and, if it is well run, a real process for killing bad ideas early. Those are the hard capabilities, and most large organisations now have them.

Adobe open sourced its idea generation programme, Kickbox, years ago precisely because idea generation was not the scarce thing.

Commercialisation is a different discipline with a different clock. It requires a positioning decision, a brand, a price, a channel, a launch date, a media plan, retail or direct commerce infrastructure, a support function, and a named person who is accountable for the revenue number.

Inside a large company, almost every one of those assets belongs to someone else. That someone else is being measured quarterly on brands that are already ten or a hundred times larger than the new one.

The result is the specific failure mode this article is about. The invention succeeds and the launch never fully happens.

What a Mondelēz International VP said on the record

Mondelēz International set up SnackFutures in 2018. It created five brands. Two of them, Dirt Kitchen and CaPao, reached US retail, a third, NoCoé, sold only in France, and Mondelēz International began phasing the created brands out at the end of 2022, as Food Dive reported in September 2023.

The reason was given plainly by Richie Gray, VP at Mondelēz International, in AgFunderNews in September 2025: "We launched a lot of products, but we struggled to really bring them to life because we rely on our business units to do that, and our business units had bigger businesses and bigger brands to focus on."

Read that sentence again as a structural description rather than as a confession. The invention step worked. Products were made and shipped to shelves.

The commercialisation step depended on units that had a rational reason to spend their attention elsewhere. Nobody behaved badly. The org chart produced the outcome.

The closure pattern is not one industry

If the problem were food and beverage, or a single management team, it would show up in one sector. It does not.

Walmart's Store No. 8 is the cleanest arc because the numbers are public at both ends. It was created in 2017 by Marc Lore, then CEO of Walmart e-commerce.

Its first launched product, Jetblack, charged $50 a month and had fewer than 1,000 customers. TechCrunch reported in February 2020, citing the Wall Street Journal, that it was losing as much as $15,000 per year per member. Walmart shut it on 21 February 2020.

The incubator itself was closed in an internal memo from CFO John David Rainey sent on 19 January 2024, with around 300 employees moved into other technology roles, as reported by Talk Business & Politics.

General Mills is the most recent and the most instructive, because the capability was not abolished, it was reabsorbed. Food Dive reported in March 2025 that G-Works was shut and new outside investments through 301 Inc were paused, with a spokesperson saying the company was "adjusting how we pursue new growth initiatives" and a Strategic Growth Office taking over.

Not every closure is a failure, and it would be dishonest to write it that way. SPACE10 ended on its own terms. Co-founder Kaave Pour said that "Space10 was never meant to last", adding that after a decade working with IKEA the team had achieved what it initially set out to accomplish, as reported by It's Nice That and other design publications in August 2023.

General Motors made a portfolio decision rather than an abandonment. It announced the closure of its Chandler, Arizona IT Innovation Center affecting around 936 employees, and senior director of corporate news relations Kevin Kelly framed it as a move "to better align our Innovation Center footprint and IT resources in the U.S.", per CIO Dive, while three other innovation centres stayed open.

The honest reading of the table is not that innovation units are a waste of money. It is that the output of an innovation unit is fragile in exactly one place, the handover. When a programme is reviewed, the question asked is what it commercialised, not what it invented.

The lab is not the variable

Amazon's hardware lab, Lab126, produced the Kindle and the Echo. It also produced the Fire Phone, on which Amazon took a $170 million charge in the third quarter of 2014, with $83 million of Fire Phone inventory still on hand at the end of that quarter, reported at the time by GeekWire from Amazon's own earnings disclosure.

Same lab, same engineers, same company, three very different commercial outcomes. The lab was not the variable.

Google Glass makes the same point more sharply. The device was built and it worked. The Explorer programme sold it to a self-selected audience at $1,500, the price TechCrunch reported in April 2013 as Google prepared to put the hardware in developers' hands, and Google ended sales of that version on 19 January 2015.

Price, audience, distribution and the management of public reaction are all launch decisions, not engineering decisions.

Xerox PARC is the oldest version of the story and gets cited so often it has lost its edge, so one line is enough. PARC produced the Alto, the graphical user interface, the mouse, Ethernet and laser printing, and Xerox built an enormous business out of the laser printer and not out of the personal computer.

The Computer History Museum's account and the standard histories put the cause in the transfer between the lab and the divisions that made and sold things. It is the same handover the modern examples keep failing at.

What the research supports, and what it does not

Start with the claim you should not use, because it is on every agency blog on this topic. There is a widely repeated line that up to 90 per cent of corporate innovation labs fail, usually attributed to Capgemini. That figure does not appear in the Capgemini report it is attributed to.

What the report does contain is more useful anyway.

Capgemini surveyed 1,700 respondents at 340 organisations, each with at least a billion dollars of annual revenue in 2016, across eight countries and five sectors in March and April 2017. It found that 87 per cent of organisations have a lab or space dedicated to innovation, but only 17 per cent carry out innovation beyond the innovation centre.

On process, the report is blunt: "Many organizations lack established processes for managing an idea through to commercialization."

On maturity, 76 per cent of respondents placed their own organisation at level 2, "Building", and 21 per cent at level 3, "Established". In the all-employee view nobody placed their organisation at level 4, which the report defines as "a consistent track record of taking ideas from inception through to significant commercial or operational impact". Among leadership respondents alone, 5 per cent did.

BCG's 2024 innovation work points the same way from a different angle: 83 per cent of companies see innovation as a top-three priority, and 3 per cent are ready to translate those priorities into results, against 20 per cent scoring as ready on the same measure in 2021. Ambition is close to universal. Readiness to convert it is not.

Steve Blank named the behaviour that fills the gap. In his 2019 piece on innovation theatre, also published by Harvard Business Review, he wrote that hackathons, design thinking classes and innovation workshops shape and build culture, but "they don't win wars, and they rarely deliver shippable/deployable product". Organisations, he argued, lack the shared beliefs, validated principles, tactics, organisation and budget to explain how and where innovation gets applied and how it relates to the rapid delivery of new product.

One more correction worth carrying into your next steering meeting. The familiar figures that 80, 85 or 95 per cent of new products fail are not supported by the literature. Castellion and Markham addressed exactly this in the Journal of Product Innovation Management in 2013, in a paper on new product failure rates and the influence of argumentum ad populum, arguing that the popular claim displaces peer-reviewed findings that are substantially lower.

If a failure rate is quoted in your business case, ask which study it comes from and what definition of failure it uses. Usually there is no answer.

The counter-example: how FirstBuild bought its own route to market

GE Appliances runs an innovation hub called FirstBuild. In 2015 it took a nugget ice maker called Opal to Indiegogo. The framing in GE's own press release was explicit, from Natarajan Venkatakrishnan, director of FirstBuild: "We developed Opal to offer an affordable nugget ice maker for home use and we are using Indiegogo as a selling platform to gauge market acceptance of the product."

That is the important sentence. The campaign was not a fundraising exercise for a company that owns appliance factories. It was a market test that produced a route to market, and it did not require the parent's normal channels to approve the product first.

FirstBuild's own product page records that Opal earned more than $2.7 million through Indiegogo and that it lived at FirstBuild for several years before becoming the founding product of GE Appliances' relaunched small appliances division.

The payoff is on the record too. Marking a decade of FirstBuild in August 2024, GE Appliances described the Opal Nugget Ice Maker as "the bestselling small appliance for GE Appliances and primary driver for GE Appliances' re-entry into the small appliances category in 2020". The same release reports more than 100 products and features developed, 37 of them graduating to GE Appliances' house of brands, a community of over 245,000 cocreators, and products reaching market or crowdfund in an average of six months.

Note what was solved there, because it is easy to draw the wrong lesson. FirstBuild did not out-argue the business units. It went around the priority contest by acquiring demand evidence and a commercial channel of its own first, then handed the parent a product with proof attached.

Evidence beats assertion in an internal review. A funded campaign with real buyers is evidence that a slide cannot match. We call this The Deposit Test, and it works the same way inside a corporation as it does for a founder: if people will not put money down for the product, the business unit is right to be sceptical, and if they will, the argument is over.

Other structural answers large companies have tried

Sony built its own crowdfunding and e-commerce platform, First Flight, launched in 2015 as an extension of its Seed Acceleration Program, so that products developed by its own employees could be validated and sold outside the company's normal channels. It launched as a Japan-only service. A major electronics manufacturer had decided that its existing route to market was the wrong one for new internal products.

Samsung's C-Lab solves the handover by letting products leave. At CES 2026 Samsung exhibited 15 C-Lab startups, and C-Lab companies won 17 CES 2026 Innovation Awards including two Best of Innovation awards, for MangoSlab and StudioLab, both of which started inside C-Lab and later spun off as independent companies, per Samsung's newsroom.

A spin-out is an admission that the parent's channels were not going to carry the product, and a deliberate answer to it.

PepsiCo took a third route. It formed The Hive in 2018 inside its North America division as a team dedicated to holding and nurturing smaller and niche brands, as Food Dive reported at the time. Our reading, which goes beyond what the reporting states, is that The Hive is this article's diagnosis implemented as an org chart: it gives small brands a commercial owner who is not being measured on the core portfolio.

None of these are cheap, and none of them are governance-free. What they have in common is that each one gave new products a commercial owner that was not competing with the core business for attention.

What to settle before you build the next thing

Corporate appetite for innovation is not shrinking. Global Corporate Venturing counted around 57 corporations creating new startup investment arms in 2025, up from 46 in 2024.

The money and the mandate exist. The commercialisation route is what keeps going missing.

Five things are worth fixing before the next prototype is finished. All of them are cheaper to settle at the start than to negotiate at the end.

Name the commercial owner of the launch, by name, with a revenue number against it. If that person does not exist, the product has already been deprioritised and nobody has said so yet.

Give the launch its own budget line inside the innovation programme, not a request into a business unit's marketing plan. A request competes. A budget commissions.

Decide the route to market before the build finishes. Direct commerce, pre-order, crowdfunding, retail or a licensing deal all imply different products, packaging and timelines, and this is the decision our strategy work exists to force early.

Define what evidence will trigger a business unit handover. A number of paid pre-orders is a threshold. "Positive internal feedback" is not.

Write down what happens if the business unit says no. A spin-out, a direct-to-consumer line, a licence or a quiet shutdown are all acceptable answers. Having no answer is how a good product becomes an orphan.

If you want the operational version of the route-to-market and evidence questions, our go-to-market strategy guide covers the sequencing, and the product launch checklist covers the execution detail in the final 90 days.

Where an outside launch partner helps, and where it does not

Start with what an agency cannot do, because a senior reader will notice the omission otherwise. An outside partner cannot fix governance. If the innovation unit has no mandate to commercialise and no budget of its own, hiring anyone, including us, produces a better marketed orphan.

The sponsor still has to own the commercial decision, and that decision has to be made by a person with the authority to make it.

What an outside partner does fix is scope and ownership. Blazon is a full-service agency, which means brand, PR, paid media, commerce and post-launch growth run inside one engagement, directed by the team that owns the product, on its own timeline. That is the specific thing an innovation unit cannot assemble internally without first winning a priority contest against brands many times its size.

It is also why our work sits with product launch rather than with campaign execution. A corporate product needs those functions moving together, not sequentially through four internal queues.

The second thing we bring is the evidence mechanism. Blazon has run more than 500 product launches and 300 crowdfunding campaigns since 2016, driving over $250M in sales, and the pre-order motion is the most direct way we know to prove demand before asking a business unit to commit shelf space, inventory and media budget.

That is what crowdfunding is for in a corporate context, and it is exactly what GE Appliances used Indiegogo for with Opal. Not every product suits it. Where it does, a funded campaign converts an internal argument into a measured fact.

Be realistic about cost. Our minimum engagement is $15,000 and full launch programmes start from $40K. A corporate launch is a multi-function programme, and pricing it as a campaign is how it ends up under-resourced and then judged for underperforming.

If you want the route to market to be external, capable and accountable to you rather than to a business unit's quarterly plan, that is what Blazon's product launch team is built for, in London and New York.

The real instruction is not "hire an agency". It is this: give the next launch a named owner, a budget and a route to market before you build it.

Every programme in the table above had the first half of that sentence. None of them had the second.

Frequently asked questions

Why do products invented inside large companies fail to launch?

Because launch capability sits outside the innovation function. Brand, PR, paid media, commerce and retail belong to business units whose targets are this year's numbers on existing lines, so a new product arrives with no P&L, no shelf space and no media budget. It is rarely rejected outright. It is deprioritised in favour of larger brands, which is what Richie Gray, VP at Mondelēz International, described when he said the company struggled to bring products to life because it relies on business units that had bigger brands to focus on.

Is the innovation lab itself the problem?

Usually not. Amazon's Lab126 produced the Kindle, the Echo and the Fire Phone, which tells you the lab is not the variable. The variable is what happens at the handover: whether the product has a commercial owner, a budget, a route to market and an agreed evidence threshold for getting business unit support. Capgemini found that 87 per cent of large organisations have an innovation space while only 17 per cent innovate beyond it, and that many organisations lack established processes for taking an idea through to commercialisation.

Do 80 or 90 per cent of new products fail?

Those figures are not supported by the peer-reviewed literature, and the related claim that up to 90 per cent of corporate innovation labs fail does not appear in the Capgemini report to which it is usually attributed. Castellion and Markham addressed the new product failure rate myth in the Journal of Product Innovation Management in 2013 and argued that the popular numbers displace substantially lower empirical findings. If a failure rate appears in a business case, ask for the study and the definition of failure.

Can crowdfunding work for a corporate product launch?

It can, when the product is physical, visual and aimed at consumers, and when the aim is demand evidence rather than capital. GE Appliances used Indiegogo for the Opal nugget ice maker and said explicitly that it was using the platform to gauge market acceptance. Opal earned more than $2.7 million on Indiegogo and GE Appliances now describes it as its bestselling small appliance and the primary driver of its re-entry into small appliances in 2020. It is not the right route for every corporate product, and it is not a substitute for a channel plan.

What does a corporate product launch programme cost?

Blazon's minimum engagement is $15,000, and full launch programmes start from $40K. The range reflects scope rather than duration: brand, PR, paid media, commerce and post-launch growth running together costs more than a single campaign, and a corporate launch that is priced as a campaign tends to be under-resourced and then blamed for underperforming.

Who should own the launch of an internally developed product?

One named person with a revenue number attached, holding a budget that does not depend on a business unit's marketing plan. If ownership is shared between the innovation unit and a business unit without a written trigger for the handover, the product will sit in the gap. Write down in advance what evidence moves it across, and what happens if the business unit declines it.

What should we settle before the next build starts?

Five things: the commercial owner, the launch budget line, the route to market, the evidence threshold that triggers business unit adoption, and what happens if the business unit says no. Settle them before the prototype is finished. Every one of them is cheaper to agree at the start of a programme than to negotiate once a finished product is looking for a home.