OCT 11, 2026 · 11 MIN READ
The decline stage is the last of the four product life cycle stages, when sales and usage fall because customers have moved to a substitute. The company's job at that point is to decide whether to harvest the product for cash, refresh it or shut it down.
In most of the decline stage examples below, the substitute is AI or a successor from the same company.
You're in the decline stage when sales or usage keep falling for several quarters while a substitute grows.
Most products slide into decline from a long maturity stage, so the first signs look like an ordinary bad quarter.
Sales. Watch units and active users before revenue, because a price rise can hide a shrinking base for a while.
Competition. The threat comes from a substitute rather than a direct rival. DVDs lost to digital copies and streaming.
Pricing. Some vendors cut prices to hold on to the base. BlackBerry's fiscal 2013 annual report blamed part of its falling revenue per user on price cuts meant to keep subscribers.
Marketing. Spend gets cut back. Netflix's 2013 annual report called marketing costs for its DVD segment immaterial.
The loudest signal is the vendor's own paperwork. An end-of-support date or a successor named in a press release tells you more than any market report.
In the decline stage, strategy moves from growth to cash and exit, and each of the 4Ps gets cut back to what the remaining customers still pay for.
Product. Stop building features. The work shrinks to security fixes and a clean exit path.
Price. Hold the price for loyal users, or charge for extended support. Microsoft lists Windows 10 Extended Security Updates for businesses at $61 per device for the first year, as of September 2026, and the price doubles every year for up to three years.
Place. Distribution narrows to existing customers. When Microsoft announced Skype's retirement, it also stopped selling paid Skype features to new customers.
Promotion. What's left of it points to the substitute. Microsoft's end-of-support page for Office 2021 carries a "Buy or try Microsoft 365" button.
The clearest recent decline stage examples are Google Assistant, Chegg, Stack Overflow's public Q&A, the Tesla Model S and Model X, Windows 10, Office 2021, the Mac Pro, Belkin Wemo and Skype. BlackBerry, DVDs and landlines are the older textbook cases.
Google launched Assistant in 2016, and in January 2020 said it helped more than 500 million people every month.
In March 2025 Google called generative AI another platform shift and said millions of people had already switched from Assistant to Gemini. Google's August 2026 announcement set September 3 as the start of Assistant's removal from phones, tablets and paired devices like Wear OS watches, with no way to switch back. Cars with Google built-in keep it for now.
Chegg's subscription services reached 8.2 million students in 2022, per that year's annual report, and fell to 6.6 million in 2024. Its Academic Services revenue was down 61% year over year in Q2 2026.
Its filings say students see generative AI products like ChatGPT as strong alternatives, and that Google's AI Overview keeps searchers on Google.
Chegg is harvesting.
A restructuring announced in October 2025 was expected to affect about 45% of its workforce, and Chegg said it would run the academic products at a much lower cost.
Chegg said in August 2026 that the next generation of Chegg, aimed at helping students get internships and then jobs, was soft launching in Q3.
Stack Overflow's question count peaked in 2016. Per the Stack Exchange API, 2,182,924 questions carry that year's date, against 1,334,845 in 2022 and 109,272 in 2025.
The company said a roughly 14% traffic drop in April 2023 was likely down to developers trying GPT-4. Its owner, Prosus, recorded a US$280 million goodwill impairment on Stack Overflow in its 2026 annual report, blaming disruption from generative AI.
Stack Overflow made AI Assist the new entry point to the public site in December 2025 and put Stack Overflow for Agents into beta in June 2026.
Stack Overflow is trying a refresh rather than a harvest.
Tesla delivered 101,420 Model S and Model X cars in 2017, and its annual report for that year said the pair made up a significant percentage of its revenue. By 2022 they were at 66,705, against 1,247,146 for Model 3 and Model Y.
On the January 28, 2026 earnings call, per the Motley Fool's transcript, Elon Musk said it was time to end both programs and that Tesla was moving into a future based on autonomy. Tesla's July 2026 update says it has decommissioned the S and X lines at Fremont and is installing Optimus robot lines in their place.
Tesla picked shutdown over harvest, and I think the factory floor was worth more to it than the tail.
Support for Windows 10 ended on October 14, 2025. It launched in July 2015 and was on more than one billion active devices by March 2020.
The substitute is Windows 11, which Microsoft said in January 2026 had reached one billion users. StatCounter's web-traffic data puts Windows 10 at 64.14% of desktop Windows usage in August 2024 and about 30% in August 2026.
Microsoft is harvesting the tail with Extended Security Updates.
Office 2021's support runs from October 5, 2021 to October 13, 2026, with no extension and no extended security updates.
Microsoft's FY2026 annual report ties its consumer Office revenue partly to the continued shift from on-premises Office to Microsoft 365 Consumer subscriptions. Those subscribers were up 7%.
In April 2026 Microsoft told organizations on Office LTSC 2021 to begin upgrading to Microsoft 365 Apps now, and named Office LTSC 2024 as the path for those not yet ready to move to the cloud.
The Mac Pro, which launched in August 2006, was still at $6,999 in March 2026 with no update since June 2023, 9to5Mac noted. A year earlier, Apple had called the new Mac Studio the most powerful Mac it had ever made.
Apple confirmed to 9to5Mac on March 26, 2026 that it was discontinuing the Mac Pro, with no plans for future Mac Pro hardware.
Apple doesn't break out Mac Pro sales, so the signal here is neglect. When a product stops getting updates while a sibling gets the attention, I assume the decision has already been made.
Belkin launched Wemo, its line of smart home accessories, in 2011. Its end-of-support notice dates the last sale of each affected model, from the Wemo Link in August 2015 to the Wemo Mini Smart Plug (WSP080) in November 2023. The iOS app got one update after February 2021.
In a notice dated July 10, 2025, Belkin said it would end support for older Wemo products effective January 31, 2026. The current version of the notice says the Wemo app, remote access and the Alexa and Google Assistant integrations will no longer function for affected products after that date.
The notice doesn't name a replacement. It only says Belkin is moving resources to the rest of its portfolio.
Microsoft agreed in May 2011 to buy Skype for $8.5 billion, citing 170 million connected users the year before. The substitute turned out to be Microsoft's own Teams.
Microsoft said in February 2025 that consumer meeting minutes in Teams had grown 4X in two years. Skype was retired on May 5, 2025, and its users could sign in to Teams Free with their Skype account and find their chats and contacts there.
BlackBerry sold about 52.3 million devices in fiscal 2011. Two years later it shipped about 28.1 million, and the company partly blamed its falling revenue and market share on buyers wanting the apps on iOS and Android.
The exit came in stages. BlackBerry said in September 2016 that it would end all internal hardware development and outsource it to partners. It set January 4, 2022 as the date its BlackBerry OS and BlackBerry 10 phones would stop working reliably, including for 9-1-1 calls.
Its 2026 annual report describes two core divisions, QNX and Secure Communications, and never mentions a smartphone.
CNBC put DVD sales at a $16.3 billion peak in 2005 and $2.2 billion by 2018. Industry group DEG says US spending on buying discs of all formats fell 9.3% in 2025, while subscription streaming took more than 92% of US home entertainment spending.
Netflix ran the textbook harvest. Its 2013 annual report said DVD memberships were declining and that it didn't plan to add resources to DVD operations, while it expected the business to keep generating significant contribution profit. It mailed its last DVDs on September 29, 2023.
DEG says 4K UHD spending rose 12% in 2025, so a premium niche can keep growing inside a declining category.
By the second half of 2025, 80.0% of US adults lived in wireless-only households and 1.0% in landline-only ones, per CDC data released in September 2026.
Revenue in AT&T's Legacy segment, voice and data over copper, fell 25.9% year over year in Q2 2026, while direct operating costs fell only 10.8%.
AT&T aims to stop serving the large majority of its copper network by the end of 2029. It expects the segment's EBITDA to turn negative after 2027 and stay negative until the network's direct costs are substantially eliminated.
A product in decline had a market and is losing it, while a failed product never found a market in the first place.
The playbooks are opposite. A failing product needs a pivot or a fast shutdown. A declining product still has paying customers, and they're owed a planned exit.
BlackBerry and 3D TV both sit on my list of famous failed products, and they land on opposite sides of this line. BlackBerry had real growth first, with the number of devices sold up 43% in fiscal 2011, so it's a decline.
3D TV is often listed as a decline, but I'd put it closer to a failure. NPD DisplaySearch had 3D TV at 25 million units in 2011.
As early as January 2011 the firm said competitive pressures were rapidly making 3D a compulsory feature.
In 2012 it found owners hadn't watched much 3D content on their sets.
I don't count a feature that came with the set and that owners barely used as one that found a market.
AI is shortening the decline stage, either by replacing a product with an AI successor, as Gemini is doing to Google Assistant, or by answering the question before the user reaches the product, as with Chegg and Stack Overflow.
DVD sales took 13 years to fall more than 86%, per CNBC. Stack Overflow's question count fell further than that in three years, from 2022 to 2025.
If your product's main job is answering a question, I'd plan the successor now, before the traffic chart makes the decision for you.
The most common decline-stage mistake is waiting too long to decide, because a product that still makes money feels safe to leave alone. Put a date on the decision and write it into your product strategy.
The other mistake is reading decline as a marketing problem. More ad spend won't bring back users who've moved to a substitute. It just raises the cost of each one you keep.
A product manager's job in the decline stage is to run a clean exit for the customers who remain while making sure the successor is ready to take them.
Change the metrics first. New sign-ups stop mattering. Track active users and the cost to serve each one, because together they tell you how long the product can still pay for itself.
Then write the notice, with the date, what stops working, what happens to customers' data and where they go next in the first paragraph, the way Skype pointed its users to Teams. Call your biggest customers before the email lands, and write every commitment down.
For a first sunset, I'd learn the basics in a product management course rather than on customers who are already leaving.
If the successor is your company's first AI product, pin its success metrics to numbers and put an evaluation plan in the scope before you commit.
RELATED READING
The product life cycle (PLC) is the series of steps through which every product goes. Product life cycle stages - Introduction, Growth, Maturity and Decline.
The maturity stage of the product life cycle is the period when sales growth slows or stops because most of the people who will buy the product already own it. Competition peaks, and the work shifts…
As a Product Manager, we don't want our products to fail. Let us analyze some of the biggest failed products in technology to see why they failed.
Choosing the right product strategy framework is very important for every product manager. We will go over some top models like Kano, OKR, Hooked, etc. The same frameworks apply when you're shaping…
You already know this - The number one skill in product management is communication. But can you really communicate like a PM? Easy guide to get started.
This is my list of the product management courses and certifications that are worth your time, updated for 2026. AI product management programs now lead the list. Once you've picked a course and…
FREQUENTLY ASKED