Adobe: 72% of shoppers install a retail app for one discount, then delete it

Adobe found that 72% of shoppers have installed a brand or retail app to claim a single discount or complete a single purchase, then removed it. The company surveyed 1,000 consumers in the United States, published the findings through its Adobe for Business blog, and circulated them to trade press on 21 August 2026. Adobe states a 95% confidence level with a margin of error of plus or minus 3 percentage points, and notes that the results are self-reported. Alongside the 72% headline, respondents reported holding an average of ten retail or brand apps, of which five had been opened in the past month and three had never been opened at all. PPC Land reported the study on 21 August 2026 and its write-up is at PPC Land.

Start with the caveat, because it changes how the number should be used

Adobe says the results are self-reported, meaning they may reflect perceptions rather than logged behaviour. That disclosure carries more weight in this study than it would in most, and PPC Land flagged why: app deletion is a measurable event on a device. It has a timestamp. It can be counted directly through panel data or through an SDK sitting inside the app. Adobe did not pair the survey with either. The study asked 1,000 people to recall something that could have been observed.

The consequence is specific rather than general. Recall questions of this shape tend to be answered by people summarising a habit rather than counting instances, so 72% is best read as the share of consumers who recognise the behaviour in themselves, not as a measured deletion rate. It is still a useful number. It is a poor number to put in a business case that needs a churn rate, and a good number to put in front of a team that believes an install is a durable asset.

The stated precision applies to the whole sample. With 1,000 respondents at 95% confidence and plus or minus 3 percentage points, two findings three points apart are not meaningfully different from each other, which matters later in this article when the notification rankings are read too closely.

Ten apps, five dormant, three never opened

The ownership figures are the most concrete part of the study. According to Adobe, customers report holding an average of ten retail or brand apps, five of which they have opened in the past month, and three of which they have never opened at all. Half the installed base is inactive within a thirty-day window. Close to a third has never returned a single session.

The three-never-opened figure is the awkward one for acquisition teams. An app that has never been opened was installed and then abandoned before it did anything, which means the install itself completed while the value exchange behind it did not. PPC Land put the implication plainly: install campaigns are budgeted as acquisition and this research prices them as rental.

Why people install

Install motivations concentrate on incentives. According to Adobe, the leading reasons for downloading a retail app are joining a loyalty or membership programme at 45%, receiving a one-time discount or promotion at 42%, and obtaining an easier shopping experience at 37%. Respondents could select more than one reason, so the figures are not shares of a single pie and should not be added together.

Two of the top three are transactional. A loyalty enrolment and a one-time discount or promotion both describe a shopper accepting an app in exchange for a benefit available at that moment. Only the third reason, an easier shopping experience, describes an expectation about the app itself. That distribution is worth holding in mind against the deletion data, because the incentive that produced the install is spent the moment it is redeemed.

Why people delete, and the two reasons that are one reason

According to Adobe, the reasons customers give for deleting retail apps are no longer needed for the original purpose at 64%, takes up too much storage at 43%, lack of ongoing relevance or utility at 43%, lack of incentives or deals at 38%, and notification fatigue at 30%.

PPC Land made two observations about that list which are worth repeating precisely. The first is that two of the five are functionally the same complaint expressed differently: no longer needed at 64% and lack of ongoing relevance at 43% both describe an app that stopped earning its place after the transaction closed. The second is that storage, at 43%, is the only reason on the list that has nothing to do with marketing decisions. A brand cannot make a phone bigger.

The ranking itself is the finding most likely to surprise a lifecycle team. Notification fatigue comes last, at 30%. In this dataset, being irrelevant beats being noisy by a wide margin, which runs against the common working assumption that over-sending is the primary churn driver. The gap between 64% and 30% is far outside the study's own error band, so unlike some of the closer comparisons in this research, that ordering is real within the sample.

What actually gets an app opened

Adobe asked which alerts are most likely to drive a retail app open. The answers were order or delivery status updates at 39%, personalised discounts or special offers at 37%, price drops on saved or viewed items at 33%, expiring rewards, points or credits at 30%, and wishlist items back in stock at 13%.

Transactional messaging leads promotional messaging, but only by two percentage points, which sits well inside the stated margin of error. Those two categories are statistically indistinguishable in this sample and should not be reported as a ranking. What does separate cleanly is the bottom of the list. Restock alerts on wishlist items collapse to 13%, less than half the rate of expiring rewards at 30%, while price drops on items a shopper has already saved or viewed hold at 33%.

Adobe also asked what keeps an app on the home screen, which works as a proxy for perceived usefulness. Frequent use dominates at 61%, ahead of ease of navigation at 24% and exclusive offers or early access to products at 23%. PPC Land noted that Adobe's wording for the second factor drifts between its blog and its media summary, appearing as ease of navigation, ease of use and user-friendliness in different places, while the figure stays at 24% in all three renderings.

The AI feature ranking is a logistics list

Adobe asked which in-app AI capabilities would make respondents more likely to keep a retail app. Order tracking and status updates led at 65%, loyalty points and rewards tracking followed at 57%, instant returns processing reached 42%, tailored product comparisons 31%, personalised restock and sale alerts 28%, and automated customer support or troubleshooting 23%.

The ordering rewards a careful read, and PPC Land supplied it. The top three items on a list of AI features are order tracking, loyalty balance tracking and returns. None of those is a generative capability in any strict sense. Each describes retrieving a status from a back-end system and presenting it clearly. The two capabilities that involve model reasoning, product comparisons and automated troubleshooting, sit at the bottom at 31% and 23%.

Adobe separately reports that 41% of respondents said built-in AI-powered post-purchase support would make them more likely to keep a retail app on their home screen, while 21% said such features would make them less likely to delete it. The company presents both numbers without reconciling them, and PPC Land noted they measure closely related dispositions at roughly a two-to-one difference. A further finding covers input methods: 33% of respondents, or one in three, said the ability to interact through voice or image search would make them more likely to keep a brand or retail app installed.

Loose ends Adobe did not close

Several parts of the publication do not line up, and PPC Land documented each of them. The Adobe for Business article carries a date stamp of 08-12-2026, which under United States convention reads as 12 August 2026, nine days before the 21 August 2026 media distribution that described the study as launching that day.

The generational breakdown is presented in the same percentage units as the all-respondent figures, which creates ambiguity. Gen X grocery and food delivery is listed at 39%, identical to the all-respondent grocery figure of 39%, and the Baby Boomer home goods figure of 6% differs from the overall 5% by a single point. Subgroup sample sizes are not disclosed, and a 1,000-person sample split across four generations produces cells where the total-level margin of plus or minus 3 percentage points becomes considerably wider.

There is also a scope question the research does not address. PPC Land reported that Adobe's published product description for Brand Concierge, effective 11 November 2025, states the product may be deployed on customer sites excluding mobile applications, while Adobe's implementation documentation describes it as embeddable across websites, mobile apps and integrated SDKs. A study about retail mobile app retention is therefore paired with a product whose licensing terms and technical documentation disagree on whether mobile apps are in scope. Product Support Agent, the other product positioned against these findings, was among six agents Adobe made generally available on 10 September 2025 on Experience Platform Agent Orchestrator, where it was described as a tool for resolving issues for Adobe's own enterprise customers.

Every stated reason for deleting a retail app

The table below lists the uninstall reasons exactly as Adobe reported them, with the share of the 1,000-respondent sample giving each. The margin of error of plus or minus 3 percentage points applies to each figure, which is why the two reasons tied at 43% should be treated as tied rather than ranked.

What a retailer actually does with this

This section is analysis and is not something Adobe recommended. The study reports what consumers said. It does not prescribe retention tactics, and nothing below should be attributed to Adobe.

The organising idea in the data is that an app has to earn its place after the transaction closes. The largest single deletion reason, at 64%, is that the app is no longer needed for the original purpose, and the third largest, at 43%, says the same thing in different words. Both describe a job that finished. An install bought with a one-time discount or promotion is therefore an app with a scheduled expiry date built into the reason it exists, unless something else takes over once the discount is spent.

The second idea is that notification volume is the wrong first thing to fix. At 30%, notification fatigue is the smallest reason on the list. A team that responds to app churn by cutting send frequency is optimising the least-cited cause while leaving the most-cited one untouched. The more defensible reading of the same data is that relevance, not restraint, is the lever. Order and delivery status at 39% and price drops on saved or viewed items at 33% are both triggered by first-party events that a retailer already holds, which means the highest-response message types do not require new data collection, only a connection between existing systems and the send.

The third idea sits under the AI ranking. Consumers put order lookup at 65%, loyalty balance retrieval at 57% and returns at 42% ahead of anything involving model reasoning. Each of those is an integration problem before it is a model problem. Whether a conversational layer improves retention depends on whether the order management system, the loyalty ledger and the returns workflow can be queried at all. Teams planning app install budgets through paid social campaigns can compare the cost of that integration work against the cost of replacing the installs it would save.

What this means for Thai marketers

This section is analysis, and it needs stating plainly that Adobe surveyed consumers in the United States and said nothing about Thailand. There is no Thai sample, no Thai breakout, and no basis for reading these percentages as local figures.

The structural difference matters more than the numbers here. Thai retail sits heavily on LINE and on marketplace apps rather than on standalone brand apps, so the literal finding, that shoppers delete a brand app after one discount, applies to a smaller slice of the market than it does in the United States. What transfers is the underlying failure, which is not really about apps. It is about what happens to a customer relationship once the transaction closes, in whatever channel the brand actually owns.

Read that way, a LINE audience is subject to the same pressure as an app icon. A shopper who joins to get a coupon has the same reason to stop reading after redeeming it as a shopper who installs an app for a discount has to delete it. The channel differs and the mechanic does not, and the response Adobe's data points to is the same: give the post-purchase phase something worth opening, starting with order status, which respondents ranked ahead of promotional messaging both as a notification trigger at 39% and as a desired AI capability at 65%.

The wider point for Thai ecommerce marketing teams is where discovery now happens. If a brand's owned app or owned chat channel is not the place a customer returns to, then the return visit is being routed through search, marketplaces and AI answer surfaces instead, which puts weight back on the organic work covered by SEO in Thailand rather than on the install budget.

Frequently asked questions

Does 72% mean most people delete every retail app they download?

No. It means 72% of the 1,000 United States consumers Adobe surveyed said they have at some point installed a brand or retail app for a single discount or a single purchase and then removed it. It describes a behaviour people recognise having done at least once, not a deletion rate applied to every install.

How much should the self-reported caveat worry me?

Enough to stop you using the figure as a measured churn rate. Adobe states the results are self-reported and may reflect perceptions rather than logged behaviour, and PPC Land noted that app deletion is a measurable device event that this study did not measure with panel or SDK telemetry. The direction of the finding is credible. The precision is not.

Is any of this about Thailand or Southeast Asia?

No, and the source did not say anything about Thailand. Adobe surveyed 1,000 consumers in the United States. Applying the percentages to a Thai audience would be inference, and the structural difference between a market built on standalone brand apps and one built on LINE and marketplace apps makes that inference weaker than usual.

Should a retailer cut push notification frequency?

The data does not point there first. Notification fatigue was cited by 30% of respondents, the lowest of the five deletion reasons, while no longer needed for the original purpose was cited by 64%. Cutting frequency addresses the smallest stated cause. That is analysis rather than an Adobe recommendation, and the study makes no recommendation on send volume.

Did Adobe publish a fix for any of this?

Adobe positioned two of its own products against the findings, Brand Concierge for conversational discovery and Product Support Agent for post-purchase service, but the study itself reports consumer responses rather than tested retention tactics. PPC Land also noted that Adobe's own product description for Brand Concierge, effective 11 November 2025, excludes mobile applications from deployment scope while its implementation documentation does not.

If your post-purchase messaging still stops at the receipt, the numbers in this study describe what happens next. A short audit of which order, loyalty and returns events are actually reachable from your send system is a cheaper starting point than another install campaign, and it is a good conversation to have before the fourth quarter fills up.