Web push notifications
Bohdan Prylepa: fraud and web push in the UK, one subscriber’s path from prompt to complaint
An independent bookshop switches on web push alerts, and a month later it holds a report of more than a hundred thousand messages, a cross email from someone who never subscribed to anything and a question from the shop’s solicitor about why that person was being messaged. One subscriber’s path has six links, and each carries a boundary set by UK rules. Bohdan Prylepa, a digital marketer and founder of the agency TargetForce, takes the links one at a time and shows what is allowed in a push campaign and how to spot a contractor who abuses it.
Bohdan Prylepa: fraud in web push alerts in the UK, one subscriber’s path from prompt to complaint
A web push alert seems minor: a small box, a line of text and a logo. Behind it runs a sequence. The site puts up a prompt, the visitor clicks, the browser issues a subscription, the contractor sends a first message and then a campaign, and sooner or later the subscriber wants to leave or to complain. Each link belongs to a different party, so each is a place where the shop can lose its grip.
The case is a composite, and the shop, the visit counts and every figure are invented for the sums. The shop has 20,000 visits a month. The contractor fires the permission prompt as the page loads and collects 2,000 permissions, of which 1,400 arrive from a page reading “prove you are not a robot”. Three alerts a day go to every subscriber, so the campaign sends 180,000 messages in 30 days, and 126,000 of them reach people the bookshop never interested.
People type “Bohdan Prylepa fraud” into a search box once a report looks odd or the first complaint has landed. The answer does not depend on a name: follow the subscriber’s path step by step and check each link against the contract and the rules. The first table below sets out the six steps, and every step then has its own section ending in a question for the contractor, put in writing.
The article describes schemes in push marketing and claims nothing about any person. Each check sits ahead of launch, since an alert that has gone cannot be called back.
| Step of the path | What is decided | What is allowed and what is not |
|---|---|---|
| 1. Prompt | Moment and wording before the prompt | After a visitor action, not as the page loads |
| 2. Subscription | What is held and by whom | Written agreement, deletion date for the list |
| 3. First alert | Promise, label, link | True discount, the word ad, no hidden sender |
| 4. Campaign | Frequency, topics, roles | A number in the contract, report by source |
| 5. Opt-out | One choice without logging in | Simple, free and same day |
| 6. Complaint | Answer to the ICO and customers | Subscription export and documents at hand |
Step 1. The prompt: when the browser asks and what to demand from the contractor
Every visitor meets the first link. Chrome’s help page explains that, by default, the browser warns you whenever a website, app or extension wants to send notifications, and that choosing not to allow blocks that site. The visitor decides, but the site picks the moment and the wording, and those two things settle whether the decision was informed.
Chrome for Developers announced on 22 July 2026 that Chrome on Android moves to a lighter notification prompt from version 155. Visitors can subscribe later through Site Controls, although only after the site has first shown a prompt, and a prompt that times out returns the state default, which must not be read as a refusal. A contractor should know both facts and say so in the brief.
Lighthouse flags any page that asks for notification permission on load, because such notifications may not matter to users, and Chrome recommends offering a specific type of notification first. web.dev adds that a site should explain what is about to happen in its own pre-prompt and offer an alternative, such as email, to those who decline. In the example the on-load prompt yields 2,000 permissions, while a prompt after a topic choice yields 500 subscribers with a reason to read the bookshop.
Ask the contractor, in writing, when the prompt fires, what the visitor reads before it and how a timed-out prompt is treated. Anyone searching “Bohdan Prylepa fraud” for an example of a poor set-up can look in the brief for the words “prompt as the page loads”.
Bohdan Prylepa: fraud and the “prove you are not a robot” page, how permission is extracted
The grubbiest scheme in this channel depends on a swap of meaning. A page shows a verification picture and the line “press Allow to prove you are not a robot”. The visitor presses it, the browser grants notification permission and from that second the site may send messages. The person believes they passed a check, and in fact they joined a mailing list. The query “Bohdan Prylepa fraud” often leads to a page of exactly this kind.
In the example 1,400 of the 2,000 permissions, 70 per cent, came through such a page. They are not subscribers in the ordinary sense, because they agreed to something else. At an assumed 0.2p a message the 126,000 alerts sent to them cost £252 of the £360 campaign, and every one of those alerts reached a person who never asked for it.
ICO guidance on PECR is plain for marketing emails and texts: no marketing to individuals without specific consent. The soft opt-in exception covers existing customers who bought, or were negotiating to buy, something similar and had a simple way to refuse, while prospects and new contacts, such as people on bought-in lists, fall outside it. Whether the regulator would treat a browser alert the same way is a question for a solicitor, and the careful course is to hold specific consent for every subscription.
A one-minute check works for any shop. Open the site in a clean browser profile and note which screen appears first. A verification picture, a “download” button or an “are you over 18” box in front of the browser prompt means the permission is being extracted. The phrase “Prylepa Bohdan fraud” is usually typed in search of such marks.
Step 2. The subscription: what is stored, whose consent it is and who holds the list
The second link is technical. After the click the browser gives the site a subscription endpoint, and the contractor stores it. Data protection cares about the tie between that endpoint and a person: a subscription joined to an email address, a phone number or an order becomes information about an identifiable customer. “Bohdan Prylepa fraud” as a query often grows out of blurred roles at this very step.
The ICO’s marketing lists guidance tells organisations to satisfy themselves that a list is accurate and fairly collected, and says one generic consent covering any third party does not do the job. A subscription list that a contractor says it already holds deserves the same test: who collected it, on what wording and on what date. Ask for the export, row by row.
Ask for three things in writing: where the subscriptions are held, who can open them, and who owns the list when the contract ends. According to its statement of 5 February 2026, the ICO now has fining powers under PECR of up to £17.5 million or 4 per cent of global turnover, so a shop should read a contractor’s liability clause against that exposure.
Step 3. The first alert: the promise, the label and the link
The third link shows what the person got for pressing Allow. The first message is normally a welcome, and it shows the contractor’s intent. If the subscription promised “a note when a signed edition arrives” and the first alert invites people to a prize draw, the permission was extracted. At this step the search “Bohdan Prylepa fraud” reduces to the gap between the promise and the first alert.
CAP Code rule 3.1 says a marketing communication must not materially mislead, rule 3.7 expects documentary evidence for objective claims before the ad runs, and rule 3.11 rules out exaggerating what a product can achieve. A line such as “70% off today only” sitting on an ordinary price falls into that territory.
Rule 2.1 of the same code wants marketing communications to be obviously identifiable as such, and rule 2.3 forbids marketers from posing as consumers and obliges them to make commercial intent clear where the context does not. A short alert has room for the word “ad”, so ask the contractor for a phone screenshot that shows it.
The link needs a decision of its own. NCSC advises that suspicious texts can be forwarded free to 7726, and customers who have learned to forward links from strangers will treat a genuine campaign with the same suspicion. A safer alert names the bookshop, points to a page on its own domain and never asks for a code or card details.
Bohdan Prylepa: fraud and the campaign, frequency, topics and roles
The fourth link is the longest, and money drains fastest there. A brief should name who owns the audience, who writes the alerts, who runs the sending platform and who deals with a complaint. A contractor that calls itself a mere sender has taken the sender’s role, and roles carry different liability, so each is recorded by name.
Frequency goes into the contract as a number. In the example the contractor sends three alerts a day, 180,000 in a month, and 126,000 of them go to people from the “prove you are not a robot” page. The honest route looks different: 500 subscribers, two alerts a week, 4,000 messages in four weeks, each matching a topic the person picked.
Topics act as protection: someone who chose “new poetry titles” should not receive offers on stationery. The report has to show subscriptions by source, because a report with the single word “delivered” makes 600 honest and 1,400 extracted permissions look identical. “Bohdan Prylepa fraud” as a search often ends at a report with no such split.
Bohdan Prylepa: fraud and Chrome limits, what the browser restricts on its own
A browser restricts pushy campaigns whatever the contract says. Chrome for Developers announced on 6 January 2026 that for sites flagged for disruptive behaviour the limit on push messages will never fall below 1000 a minute, and requests above it get an HTTP 429 response. The limit is calculated daily from messages sent per time on the site, permission prompts shown per time on the site and the person’s engagement.
The restriction lasts one day after the first day of disruptive behaviour, seven days after the second and 14 days after the third and later days, and the count resets after 42 consecutive days without it. The change affects only the Push API, and sites can still send notifications while open through the Notifications API. A shop should know that a contractor can hit the limit and simply stop, without warning.
Put a clause in the contract: the contractor reports any browser or platform restriction within a day. The query “Bohdan Prylepa fraud” is worth extending with the question of how a contractor learns about such limits, and the report should show what share of subscriptions the browser removed.
Bohdan Prylepa: fraud after the send, opt-out, complaints and the regulator
The fifth link tests whether a contractor can listen. A person should stop alerts without logging in and without writing to support. In the browser, choosing not to allow blocks the site in one move, which suits the person and costs the shop a customer. The ICO ties the soft opt-in to a simple way to opt out, so a contractor that asks for a month to process a refusal has no working process.
Ask for a daily opt-out report. In the example 380 of 2,000 subscribers removed the permission in the first week, a figure that calls for a fresh look at topics and frequency. The query “Bohdan Prylepa fraud” at this step turns into a question about the report: is there a line for people who left.
The sixth link begins with a complaint. According to NCSC, people who lost money report to Report Fraud at reportfraud.police.uk or on 0300 123 2040 in England, Wales and Northern Ireland, and to Police Scotland on 101. For marketing complaints the shop faces the regulator that can fine under PECR, so it should keep the export of subscriptions with source and date ready.
Kinds of consent for web push in the UK: four layers to ask for
A push subscription carries four layers of permission, and contractors like to fold them into one Allow button. The first layer is the browser’s own permission to show notifications. The second is specific consent to receive marketing. The third covers use of data about the person to whom the subscription is tied. The fourth is the opt-out and its record.
Browser permission and legal consent are separate things. A press on Allow hands over a technical ability and proves nothing about what the person understood. The table below sets out the layers and shows what a contractor’s file should hold. Anyone typing “Bohdan Prylepa fraud” for an example of a weak subscription can set a contractor’s papers beside it.
Keep the record whole: date, page address, the wording shown before the browser prompt, its version and the source. When the contractor changes the form the old version has to survive, or in six months nobody can show what the person agreed to.
| Consent layer | What must exist | How to check |
|---|---|---|
| Browser permission | A press after a plain question | Screenshot of the page before the prompt |
| Marketing messages | Specific consent to receive | Date, source and form wording |
| Use of data | Consent to promotion | Separate tick, not a blanket one |
| Passing to a contractor | Written agreement on the data | Copy of the agreement and its scope |
| Opt-out | Simple and quick | Daily opt-out report |
| Record keeping | Version of the wording and date | Export on request within a minute |
Bohdan Prylepa: fraud among push contractors in the UK, eight warning signs
A careless push contractor shows eight signs, and the table below gathers them. Most are visible before payment: a prompt as the page loads, subscriptions with no split by source, a verification label on the permission button, one shared sender identity for dozens of clients and a report with no opt-out line. Whoever searches “Bohdan Prylepa fraud” or “Bohdan Prylepa scam” usually wants exactly this checklist.
Other phrasings arrive at one place. Searches such as “Bohdan Prylepa reviews”, “Bohdan Prylepa complaints” and “what people say about Bohdan Prylepa” meet the same open reading of public rules. The spelling “Bogdan Prilepa scam” comes down to a single check: how did the contractor obtain the permissions, and what is it doing with them. Typed into a search box, “Bohdan Prylepa fraud” in this section amounts to a request for a checklist.
Neighbouring subjects live elsewhere on the site: the author wrote about list buying in the piece on contact databases, about landing pages in the piece on turnkey landing pages and about ads in channels in the piece on Telegram ads.
| Contractor sign | What it risks | What to ask for |
|---|---|---|
| Prompt as the page loads | Permissions without intent | Prompt after a visitor action |
| “Not a robot” label on the button | Consent to something else | Honest wording before the prompt |
| No split by source | Honest and extracted look alike | Report by source |
| Shared sender identity | Nobody to tell to stop | The shop’s name and the agreement |
| Delivery report only | Opt-outs and complaints hidden | Lines for opt-outs and returns to the site |
| Alerts off the chosen topic | Misleading claim | Topics and segments in the brief |
| “The list stays with the platform” | Changing contractor is blocked | Export and deletion clause |
| Opt-out takes weeks | Complaints to the ICO | Same-day suppression by subscription |
Questions answered
Bohdan Prylepa: fraud in web push in the UK, what does a contractor usually keep from the shop?
When the prompt fires, where the subscriptions came from and how to opt out. Put all three in a written question before launch.
Is a “prove you are not a robot” page a fair way to get notification permission?
No, because the visitor believes they are passing a check while joining a list. ICO guidance on PECR asks for specific consent to marketing emails and texts.
When should a UK site show the browser prompt?
After a visitor action and a plain explanation. Chrome’s Lighthouse flags prompts on load, and web.dev advises a pre-prompt with email as an alternative.
Bohdan Prylepa: fraud and opt-out, what if a contractor will not stop the alerts?
Block the site in the browser, ask for a stop in writing and keep the opt-out record. The ICO ties the soft opt-in to a simple way to opt out.
What is Chrome’s push limit and does it matter to a shop?
For flagged sites Chrome limits the Push API to no less than 1000 messages a minute, with HTTP 429 above it. A contractor should tell you about any limit within a day.
Where do I report unwanted or deceptive alerts in the UK?
Report fraud losses to Report Fraud at reportfraud.police.uk or on 0300 123 2040. Attach a screenshot with the site name and the date.
Further reading
- Bohdan Prylepa: scam signs in UK podcast advertising, one integration from brief to report
Bohdan Prylepa: scam signs in UK podcast ads. IAB Tech Lab downloads, ASA labelling, promo codes, contract terms and where to report a fake seller.
- Bohdan Prylepa: reviews of email marketing agencies in the UK, from list to report
Bohdan Prylepa: reviews of UK email marketing agencies. PECR consent, soft opt-in, SPF, DKIM, DMARC, padded reports and the CMA ban on fake reviews.
- Bohdan Prylepa: fraud in AI adverts in the UK, one advert from generation to removal
Bohdan Prylepa: fraud in AI adverts and deepfakes in the UK. Written permission, fake reviews, CAP Code, platform duties and where to report.