How to Tell Customers Their Order Is Ready
A telephone call reminder produced a 61.5% response rate against 56.7% for a text message (BJOG, 2025). That is the only randomised UK trial found that compares two notification channels head to head. It measured whether women aged 60 to 79 returned a urine sample for HPV screening, not whether a customer collects a ready order.
That gap is the closest measured evidence in this whole field, and it still is not about a waiting customer. Seven routes exist to tell someone a job is done, across four families: a voice, a screen or a pager, a text or an app alert, and email. Four pages ranking for this question were checked here, and each mostly explains how to use a single method.
What decides whether any of them fires is different: a permission granted, a number kept accurate, a sender authenticated, or a customer standing close enough to hear. This article sorts the seven by what each one requires, on both sides, before a single message goes out.
Key Takeaways
- The only in-window randomised comparison of two notification channels found a 61.5% response rate for a phone call against 56.7% for a text (BJOG, 2025). The setting was a cervical screening trial, not an order-ready alert.
- A newly installed Android 13 app cannot send a notification until the customer taps Allow (Android Developers, 2026). An app already carried through a device upgrade is usually pre-granted the permission and never sees the prompt.
- Google asks every sender for SPF or DKIM. A sender pushing 5,000 or more messages a day to Gmail accounts needs both, plus DMARC and a one-click unsubscribe link (Google Workspace Admin Help, 2024).
- Across four competitor pages fetched in full for this article, not one used the words permission, consent, opt-in, unsubscribe, SPF or DKIM.
The comparison that gets published is not the one that decides it
Search this question and the results agree with each other before they disagree with reality. Four pages that rank for it were checked for this article: SimpleTexting, Textedly, OrderReady and Shipmondo. Two are SMS template pages, one is a screen-display product’s homepage, and one is a shipping platform’s blog post. None of them compares the routes side by side, and none asks what has to already be true before a message can be sent.
All four were fetched in full on 30 August 2026 and every character of visible text on them was counted, navigation and footers included. Between them they ran past 35,000 characters and carried no statistic at all. The only number any of them printed was a 10% discount inside a sample text message, a promotional line in a template rather than a claim about the channel.
None of the four used the words permission, opt-in, consent, unsubscribe, SPF or DKIM even once. That is a search for six words on four pages, not a verdict on vendor pages in general. Textedly’s page prints “Text STOP to end” four times in its sample messages, so the mechanism is there where the vocabulary is not. Most of what ranks for this question is published by a company selling one of the methods it describes.
That gap is this article’s reason to exist. A cost comparison answers what a venue would pay if every method already worked. A precondition comparison asks whether it works at all, this week, on the hardware and the customer list a venue already has. The same four choices weighed inside a restaurant cover the cost side in full. What follows starts one step earlier: what a business and a customer each need to have in place first.
What does each method actually need before it works?
Seven routes exist to tell a waiting customer their order is ready. Each one needs something specific from the business, and something different from the customer, before a single message can land. A price or a feature list says nothing about either requirement.
A name called out
A called name needs a room quiet enough to carry it and a customer facing the right way to notice it. Nothing has to be issued, charged or granted for this route: it works the instant a member of staff speaks. What decides whether it lands is audibility and attention, not hardware or permissions. Whether a called name is heard over a room, and what beats it works through the mechanism in full.
A number on a screen
A screen or number board needs a sightline from wherever people end up waiting, not from where a floor plan assumed they would stand. Not every customer can see a screen from every spot in a room. A screen a customer cannot see delivers nothing, however bright it is. No hardware has to be handed to anyone, but the room’s own layout decides whether the precondition is met at all.
A pager handed over
A pager needs hardware issued at the counter, carried within range of the transmitter, and handed back once the order is collected. None of that depends on a customer’s own phone, battery or signal. It depends entirely on the venue’s own kit staying charged, in range and accounted for. What is inside a pager system, and what decides its range covers the hardware in full. This article prints no prices and no range figures; the precondition here is simpler: a physical object issued, and returned.
A text to a number you hold
A text needs a phone number, and a phone number is not free to hold. Collecting one turns a single transaction into a record the business now has to keep accurate and eventually delete. The duties that arrive with the number itself covers what holding it obliges a business to do. Sending the text also costs money per message. This article prints no figure for that; what a segment of SMS is billed at in the UK covers it separately.
A notification to an app
An app notification needs the operating system’s permission, and that permission is not automatic. On Android 13 and higher, a newly installed app’s notifications are off by default until the customer taps Allow (Android Developers, 2026). An app already installed before the device upgraded is usually pre-granted the same permission and never sees the prompt.
Apple sets the same condition on iOS: an app must ask before it can show a notification. The system prompts only once, recording whatever the customer answers from then on (Apple Developer Documentation, 2026). Apple also offers a quieter trial mode that needs no permission first. Those alerts make no sound, show no banner and never reach the lock screen. Either way, the precondition is a permission already granted, not a message already sent.
| Route | What the business needs | What the customer needs |
|---|---|---|
| A name called out | A member of staff and a quiet enough room | To be within earshot and facing the right way |
| A number on a screen | A screen with a clear sightline from the waiting area | A line of sight to that screen |
| A pager handed over | Hardware issued, charged and in range, then collected back | To carry it and hand it back, nothing else |
| A text to a number you hold | A phone number collected and kept accurate | A phone able to receive a text |
| A notification to an app | The app installed, and the OS permission already granted | To have granted that permission when asked |
| A WhatsApp message | Their opt-in permission, and in most cases their mobile number | WhatsApp installed on their phone |
| An email | A sending domain authenticated under the rules below | An inbox that does not filter it as spam |
Only one of these has been tested head to head, and not on customers
The Catch-Up Screen study is the only randomised UK trial found. It pits two notification channels against each other and measures who responded (BJOG, 2025). Three GP practices in Hull and Manchester invited 3,074 women aged 60 to 79 to return a urine sample for HPV screening between January and December 2024. Of those invited, 1,816 returned a sample, 59% of the total (BJOG, 2025). Each woman was randomised to a reminder by telephone call or by text message.
The telephone arm scored higher. The paper’s own words:
“The response rate was higher among those randomly allocated to receive a telephone call reminder (61.5%) compared to those randomised to receive a text message (56.7%, OR = 1.22, 95% CI: 1.06-1.41, p = 0.007)” (Rake C. et al., BJOG, 2025)
That is a 4.8 percentage point gap between the two rates, calculated directly from them rather than from the paper’s own rounded description elsewhere in the text.
The comparison is not one call against one text, either. Up to three telephone calls were attempted for each participant in the phone arm (BJOG, 2025). Weeks later, a final text went to non-responders in both arms. So the phone arm’s higher rate rests on more attempts at contact, not on the channel alone. The authors are candid about what that still cost them:
“Our ITT analysis compared a text and telephone call reminder, but 36% of the telephone arm could not be reached, so we have underestimated the potential effect.” (BJOG, 2025)
“Text messages were usually sent out earlier than telephone reminders, reflecting the time required to achieve successful participant response by telephone. A per-protocol analysis was therefore not done because 15% of participants in the telephone reminder arm had already returned their sample before they could be contacted.” (BJOG, 2025)
None of this is a study about a waiting customer. It measured whether women aged 60 to 79 in two English cities returned a screening sample. The phone arm had more contact attempts than the text arm, and non-responders in both got a final text. It is still the nearest thing to a controlled, in-window comparison this article found. Its own authors could not reach more than a third of the people they tried to call.
A phone in a pocket is not one channel, it is four
“Send it to their phone” hides four different methods behind one phrase, and each has its own requirement. A text arrives through the mobile network and needs a number the business has collected and kept accurate. A push notification arrives through an app already installed, and needs the operating system permission covered above.
A WhatsApp message needs the customer to already have WhatsApp installed. On the business side, WhatsApp’s own policy sets the conditions. It allows contact only if “they have given you their mobile phone number” and “you have received opt-in permission from the recipient” (WhatsApp). Whether WhatsApp can carry these updates covers what that asks of both sides. A fourth route needs no number and no app at all: a page opened by scanning a code.
That fourth route fails on a different precondition than the other three. The code has to land on the customer’s phone, usually at the counter, and that phone must be charged, online and have a working camera. Miss that moment, and the code has to travel to the customer somehow. That means a number or an app, which is exactly what the other three routes need.
Upd8All is one working example of that fourth route. There are no accounts and no sign-up, for either side. No phone numbers or email addresses are exchanged between the business and the customer. Pairing happens by scanning a QR code. The customer does not need the app: scanning the QR code opens the updates in their browser through Upd8 Go.
The trade-off sits on the other side of the same design. Channels are one-way and messages are capped at 40 characters. So a customer who wants to reply cannot use the channel to do it. Sorting out a mistake still means walking back to the counter and finding a member of staff.
Email arrives only if the sending side is set up for it
Email is the one route where the sending side, not the customer’s device, decides whether the message arrives at all. Google’s own guidance for Gmail states two different rule sets depending on volume. Mixing them up is the easiest way to over-comply or under-comply (Google Workspace Admin Help, 2024).
| Requirement | All senders | Senders of 5,000+ messages a day to Gmail accounts |
|---|---|---|
| Authentication | “Set up SPF or DKIM email authentication for your sending domains.” | “Set up SPF and DKIM email authentication for your domain.” |
| DMARC | Not required | “Set up DMARC email authentication for your sending domain. Your DMARC enforcement policy can be set to none.” |
| Connection | “Use a TLS connection for transmitting email.” | Same requirement |
| Spam rate | “Keep spam rates reported in Postmaster Tools below 0.3%.” | “Keep spam rates reported in Postmaster Tools below 0.30%.” (the same threshold, printed in the other list) |
| Unsubscribe | Not required | “Marketing messages and subscribed messages must support one-click unsubscribe, and include a clearly visible unsubscribe link in the message body.” |
Source: Google Workspace Admin Help, “Email sender guidelines”, retrieved 2026-08-30.
A venue emailing order-ready alerts to its own customers is almost certainly not sending 5,000 messages a day to Gmail accounts. The one-click unsubscribe requirement sits in the bulk-sender list only. It does not bind an ordinary café or garage sending a handful of alerts a shift. The all-senders list still applies to everyone, whatever the volume.
Google lists the all-senders requirements as a set, not a sequence. Four of them cover the sending setup itself:
- Set up SPF or DKIM for the sending domain.
- Confirm the domain has valid forward and reverse DNS records.
- Send email over a TLS connection.
- Watch the spam rate reported in Postmaster Tools. The requirement is to stay below 0.3%, and Google’s monitoring advice goes further: keep it under 0.10% (Google Workspace Admin Help, 2024).
None of that guarantees a message is read once it arrives. Google states plainly that it does not track open rates at all (Google Workspace Admin Help, 2024). So an email open-rate figure, whatever else it counts, is not counting anything Gmail itself measures.
Which of them lets the customer leave?
A called name, a screen and a pager all share one requirement that none of the phone-based routes have. The customer has to stay inside the venue, within reach of the method itself, until it fires. A text, an app alert, a WhatsApp message or an email all reach a device rather than a place. Each one goes wherever the customer has gone in the meantime.
That is a property of the mechanism, not a claim about how far any signal travels or how many customers wander off. The point here is narrower: three of the seven routes need a body in the room, and four do not.
Choosing between them is really a sequence of questions, asked in this order:
- Does the venue have a genuine reason to let customers leave and come back? A car booked in for an MOT is a different kind of wait from a coffee order.
- Does the business already hold, or want to hold, a phone number or an email address?
- Can every customer see a screen or hear a call from wherever the venue lets them wait?
- Does anyone need to reply, or is a one-way alert enough on its own?
The answers rule some routes out before cost or hardware ever come into it.
Every one of them fails quietly
Every route in this article can fail without producing a complaint, a refund request or anything else that shows up in the numbers a venue already tracks. A missed name, an unseen screen, a flat pager, a blocked notification and a spam-filtered email all look identical from behind the counter: silence.
| Route | How it fails | Does anyone find out? |
|---|---|---|
| A name called out | Too quiet, or the customer was not facing the right way | Only if the customer eventually asks |
| A number on a screen | No sightline from where the customer is standing | Only if the customer eventually asks |
| A pager handed over | Battery flat, or the customer wandered out of range | Sometimes, on models that report receipt |
| A text to a number you hold | Wrong or dead number, or the message never sends | Sometimes, if the platform reports a failed status |
| A notification to an app | Permission never granted, or the OS delayed delivery | Rarely; the app has no way to know the alert was seen |
| A WhatsApp message | Number not saved, or the message left unread | Rarely, beyond a delivery tick |
| An email | Filtered as spam before it reaches an inbox | Almost never; the sender gets no bounce for this |
Every one of these failures shares the same property. None generates a complaint on its own, because the customer who missed the alert has no way of knowing one was sent. What reaches a venue’s own records is only the customer who happened to ask.
Monday morning, with nothing new installed
A venue weighing these seven routes rarely needs to rank them all. The shortest path is asking which preconditions are already true today, without a single new install, grant, handover or signed-up account. A called name needs nothing but a quiet enough room, which most counters already have or already lack. A screen needs a sightline the room’s own layout already decides. A pager needs hardware that either exists in a drawer or does not.
The phone-based routes ask more of the customer before anything happens: a permission granted, a number handed over, an app already on the device. Those are real advantages once they hold, and real gaps when they do not.
The honest test for Monday morning is not which method wins on paper. It is which one could start working with nothing new installed on either side of the counter, using only what is already true today. What the message itself should say once a route is chosen is not covered here. How to write an update customers read covers the wording, once the shortlist is settled.
Frequently Asked Questions
Do you need a customer’s phone number to send them an order-ready alert?
No. A called name, a screen, a pager and a scanned code all reach a customer without the business collecting anything at all. Among the phone-based routes, a text needs a number and an app notification needs the app already installed with its permission granted. Under WhatsApp’s own policy, a WhatsApp message needs the number too, plus the customer’s permission to message them.
Why do app notifications sometimes never arrive?
Usually because the permission was never granted, invisibly to the business. On a newly installed Android 13 or higher app, notifications stay switched off until the customer taps to allow them (Android Developers, 2026). iOS asks once, and that choice decides every alert that follows (Apple Developer Documentation, 2026). An app can check whether it still holds permission (Apple Developer Documentation, 2026), but no individual blocked alert announces itself to the counter.
Is a text more reliable than a phone call?
The only measured comparison found was not about a waiting customer at all. In a UK screening trial, a phone call reminder produced a 61.5% response rate against 56.7% for a text (BJOG, 2025). The phone arm got up to three call attempts each, and 36% of it was never reached. Its subjects were women aged 60 to 79 posting a sample back for HPV testing, rather than anyone standing at a counter.
Sources
- Android Developers, “Notification runtime permission,” retrieved 2026-08-30, https://developer.android.com/develop/ui/views/notifications/notification-permission
- Apple Developer Documentation, “Asking permission to use notifications,” retrieved 2026-08-30, https://developer.apple.com/documentation/usernotifications/asking-permission-to-use-notifications
- Google Workspace Admin Help, “Email sender guidelines,” retrieved 2026-08-30, https://support.google.com/a/answer/81126
- Rake C. et al., “The Use of First-Void Urine to Screen Women Aged 60-79 for HPV in the UK: The Catch-Up Screen Study,” BJOG: An International Journal of Obstetrics and Gynaecology, retrieved 2026-08-30, https://doi.org/10.1111/1471-0528.70115
- SimpleTexting, “How to text customers when their order or table is ready,” retrieved 2026-08-30, https://simpletexting.com/blog/send-a-text-message-when-customers-food-is-ready/
- WhatsApp for Business, “Business Policy,” retrieved 2026-08-30, https://www.whatsapp.com/legal/business-policy