The classic risks of launching a new product still apply. Weak demand, bad pricing, and a late supply chain can sink anything. But in 2026, a growing share of launch failures happens after the product is designed and built, in places many teams treat as admin: the product page, the customs declaration, the sustainability claim on the box. This article covers the risks that changed this year and what to do about each one.
Key Takeaways
- From 27 September 2026, vague environmental claims such as "eco-friendly" become a banned practice across the EU, with no transition period for products already on the market.
- US tariff rules changed legal basis three times between February and July 2026. Landed cost now needs a scenario range at launch.
- AI shopping assistants send fast-growing, high-converting traffic to retailers, but they can only recommend what they can read. Product pages scored worst in Adobe's machine readability benchmark.
- Most new product launch risks in 2026 trace back to one operational weakness: product data that is incomplete, inconsistent across channels, or not linked to evidence.
The "Most New Products Fail" Statistic Is Weak Evidence
Launch decks love the line that 80, 90, or 95% of new products fail. The number circulates without a traceable study behind it. Even NielsenIQ, whose BASES unit forecasts launches for consumer goods companies, calls the 90% figure a regularly used statistic and does not present it as its own finding. What the same NielsenIQ release does report is more useful: more than half of new products typically decline in their second year on the market.
NielsenIQ also states that a binary success or fail framework does not work for measuring innovation, because retailers, manufacturers and investors judge launches by different metrics. So define success before launch, per stakeholder. A product that grows the category for a retailer can still cannibalize your own range. NielsenIQ puts a number on that risk: 1 in 4 product launches does not invest in ways that drive growth and ends up shrinking the parent brand.
The practical fix is early measurement. NielsenIQ's multi-year analysis points to weeks 4, 8, and 13 as the checkpoints where performance data tells you whether a launch is on track. Put those dates in the launch plan with owners and thresholds. A decision at week 8 is cheaper than a write-off in year two.
Regulatory Risk Moved Onto The Product Page
This is the largest change for 2026, and it hits EU sellers and anyone exporting to the EU. Several rules now regulate what a product listing says, what data sits behind it, and whether that data can be produced on request. None of them is about product design. All of them can stop a launch.
Green Claims: The EU Ban Starts On 27 September 2026
The Directive on Empowering Consumers for the Green Transition (ECGT, Directive (EU) 2024/825) amends EU consumer law and applies from 27 September 2026. It bans generic environmental claims such as "sustainable", "green" or "energy efficient" unless the trader can show recognized excellent environmental performance, for example through the EU Ecolabel or top energy efficiency class. Self-created sustainability labels are banned too. For future claims and labels, third-party verification becomes necessary, and fines can reach 4% of a trader's annual turnover in the relevant member state.
From 27 September 2026, environmental claims and sustainability labels must comply in B2C contexts, including for products already on the market and "old stock". There is no grace period for packaging that was printed last year.
The European Commission states this in its official ECGT Q&A. National consumer authorities in the CPC Network added a common position in June 2026: they expect traders to take immediate good-faith steps on old stock, such as removing non-compliant claims promptly.
For a product launching in Q4 2026, this creates a specific risk. Claims get written by marketing, approved once, then copied into packaging files, marketplace listings, retailer feeds, and sales decks. When legal changes one claim, the old version survives in four places. The ECGT also adds pre-contractual information duties on durability and repair, so the product data model needs fields that many catalogs simply don't have yet.
Product Safety: GPSR Made Online Listings A Compliance Document
The General Product Safety Regulation (EU) 2023/988 has applied since 13 December 2024. Article 19 requires distance sales offers to show specific traceability information in the listing itself: manufacturer name and contact details, the EU responsible person for non-EU manufacturers, product identifiers, pictures, and warnings or safety information in the language of the country where the product is sold.
Enforcement is rising. In 2025, the EU's Safety Gate system recorded 4,671 alerts, up 13% and the highest level on record, with follow-up actions up 35%. Those actions include removing listings from online marketplaces. By the end of 2025, more than 1,200 marketplaces had registered in the Safety Gate Portal, so a delisting request has a direct route. Nearly three-quarters of notified products came from outside the EU, with China the origin of 2,006 notifications.
A missing warning translation or an absent responsible person is a small data gap. On a marketplace, it can mean a blocked listing in launch week.
Digital Product Passports: Batteries First, Infrastructure Already Live
The Ecodesign for Sustainable Products Regulation (ESPR) makes digital product passports mandatory category by category. The first binding date is 18 February 2027, when the passport becomes mandatory for EVs, light means of transport and industrial batteries. The Commission plans the delegated act for iron and steel in Q4 2026, and construction products follow in 2027. The central EU registry opened on 20 July 2026.
If your 2026 launch contains a covered battery and stays on the market into 2027, the passport is part of the launch scope. The Commission's battery passport data-point list has already been revised as recently as August 2026. Expect the data model to keep moving. For other categories, the risk is quieter. Products launched now will still be on sale when their category's delegated act applies, and suppliers asked for material data in 2028 may no longer hold it.
Deforestation Regulation: 30 December 2026
The EU Deforestation Regulation (EUDR) applies from 30 December 2026 for large and medium-sized operators and from 30 June 2027 for most micro and small operators. It covers cattle, cocoa, coffee, palm oil, rubber, soy and wood, plus derived products. After two postponements, the Commission's May 2026 simplification package kept the date. A product that uses leather, rubber components, paper packaging, or wooden parts needs plot-level geolocation data from the supply chain before the product enters the EU.
Tariffs And Landed Cost No Longer Hold Still
US trade policy changed its legal foundation three times in 2026. The Supreme Court struck down tariffs imposed under IEEPA on 20 February. A global 10% surcharge under Section 122 replaced them on 24 February and expired after its 150-day statutory limit. On 24 July, new Section 301 tariffs of 10% or 12.5% on imports from 60 economies took effect within hours of being announced. Canadian goods face a separate additional duty under Section 338 since August 2026.
For many products, the headline rate barely moved. The legal basis did, and that matters for pricing contracts. Roughly $166 billion in IEEPA duties is moving through a refund process, and disputes over who gets refunds between importers, downstream buyers and end users are already emerging, including class actions. If your launch price includes a tariff surcharge, write down in the contract who keeps a refund.
The EU side changed too. From 1 July 2026, the EU abolished the €150 customs duty exemption for low-value parcels and applies a temporary €3 duty per item until 1 July 2028. The €3 sounds small. It applies per item based on tariff classification, so a bundle of three differently classified accessories costs €9 in duty. A Union handling fee on small parcels is due by 1 November 2026, and product identifier data becomes mandatory in declarations from the same date.
Direct-to-consumer launches shipping into the EU from abroad need this data correct per SKU before the first order:
- HS classification for each component that could be declared as a separate item, because bundle design now changes duty cost.
- Country of origin and a product identifier that matches what the marketplace listing shows.
- Customs value logic that survives an audit, since the reform targets undervaluation and split shipments.
Build the launch price on a range of landed-cost scenarios. A single-number cost model from Q1 would have been wrong three times by August.
AI Shopping Assistants Became A Launch Channel
Adobe Analytics, which tracks more than 1 trillion visits to US retail sites, reported that traffic from AI sources grew 393% year over year in Q1 2026. In March 2026, that traffic converted 42% better than non-AI sources. A year earlier, it converted 38% worse. By May 2026, AI traffic to retail sites was up 138% year over year and converted 54% better than other sources.
The same report measured how much page content large language models can actually read. Homepages averaged 75%. Individual product pages came last.
Adobe found that US retail product pages averaged a 66% machine readability score, meaning about a third of their content is invisible to the AI tools now recommending products.
A new product has no reviews, no sales history, and no backlinks. An AI assistant deciding whether to recommend it relies on whatever structured facts it can extract: dimensions, materials, compatibility, certifications, use cases. If those sit in an image, a PDF datasheet, or a JavaScript tab, the product may not exist for that assistant. Established competitors with complete attribute data win the comparison by default.
This makes attribute completeness a discovery issue, and it rewards the same discipline that EU regulation now demands. Specific, verifiable, structured facts work for both a market surveillance officer and a language model. Vague claims fail with both.
Inconsistent Product Data Turns Into Returns
The US National Retail Federation estimates that retailers will see $849.9 billion in returns in 2025, 15.8% of sales, and 19.3% of online sales. Not all of that is a data problem. Bracketing and fraud play a role. But product content is a controllable share. In Syndigo's 2025 survey of more than 8,500 consumers, 44% had abandoned a purchase because of insufficient product information, and 75% formed a negative opinion of a brand after seeing incomplete or inaccurate information.
For a new product, early returns carry extra weight. The first reviews shape the rating for months, and a return reason like "not as described" often lands in a public review. Launch-week listings on six channels built from six spreadsheets tend to disagree on at least one spec.Customers notice the wrong one.
Where A PIM System Fits And Where It Doesn't
Most risks above share a root cause. Product information lives in too many places, and nobody can say with confidence which version is current or where each claim appears. A Product Information Management (PIM) system addresses exactly that layer: one repository for product attributes, assets and documents, with rules for completeness and controlled distribution to channels.
Our customers turn to us with a familiar pre-launch setup. Technical attributes sit in the ERP, marketing copy in a shared drive, certificates as PDFs in someone's inbox, and each marketplace gets its own export spreadsheet. The problem appears late: a retailer rejects a listing for a missing attribute, or a revised safety warning never reaches one channel. In projects we implemented with AtroPIM, the first step was modeling the data. We defined the attributes each channel requires, the fields that must be filled before a product may be published, and the certificates that belong to each SKU. Completeness rules then block a product from export until the required fields exist. That doesn't make the data correct. It makes gaps visible before launch day instead of after it.
The green claims deadline produces a similar request from manufacturers: find every place a claim appears. When claims are stored as attributes and linked to their evidence documents, that becomes a filter query. Without it, it becomes an audit of every channel by hand. The same structure prepares for the digital product passport, where regulators expect structured, typed fields. A folder of PDFs won't map to them. We wrote a separate guide on preparing product data for the digital product passport.
The limits are worth stating plainly. A PIM does not forecast demand, set tariffs, or verify that a claim is true. It needs named data owners, or it becomes one more place with outdated data. And an open-source PIM such as AtroPIM moves responsibility for hosting and maintenance to you or a support partner. For a company with 50 SKUs and one sales channel, a disciplined spreadsheet may be enough. The case for a PIM grows with SKU count, channel count, and regulatory exposure, and in 2026 the last factor grew for almost everyone selling into the EU.
A Pre-Launch Risk Check For 2026
Run these checks before committing to a launch date:
- Inventory every environmental claim on packaging, listings and sales material, and match each one to evidence or remove it before 27 September 2026 enforcement reaches your channels.
- Confirm GPSR listing data for every EU channel: manufacturer and responsible person contacts, identifiers and translated warnings.
- Check whether the product contains a battery in scope for February 2027 or EUDR commodities for December 2026, and request supplier data now.
- Price the launch on at least two landed-cost scenarios, and state tariff refund ownership in customer contracts.
- Review bundle composition against the EU €3 per-item duty if you ship low-value parcels into the EU.
- Test the product page for machine readability: key specs as text and structured data, not only in images or PDFs.
- Set week 4, 8, and 13 checkpoints with thresholds that trigger a price, channel, or range decision.