Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Mobile & App Marketing › App Marketing

Web-to-App Journeys

How to move website visitors and customers into an app: where app prompts belong, links to the matching screen, signed-in continuity, and measuring the switch.
Guideintermediate11 min readUpdated

Who to move into the app, and why

Not every visitor belongs in the app. A person reading one article from a search result, buying a one-off gift or checking opening hours is best served by the page they are on, and an app prompt is a delay. The people worth moving are the ones for whom the app is genuinely better: repeat buyers whose saved details and order history live in the app, account holders who return often, subscribers who use the service weekly, people who have saved, followed or started something they will want to come back to.

Write the group down as a definition before building anything: signed-in customers with more than one order, readers who have saved content, members on a plan. That definition decides where prompts appear, which customers the emails target, and which comparison measures success. If the business is still deciding whether it needs an app at all, the guide to choosing between an app and the mobile website comes first; this guide assumes both exist and the website will keep serving everyone else.

The reason to move people should be stated from the customer's side. "The app keeps your list, works in the shop with no signal and tells you when your order ships" is a reason. "We want more app users" is not, and customers can tell which one a prompt is serving.

Where app prompts belong and where they harm the visit

A web-to-app prompt competes with the content for the visitor's attention, so placement is the main decision.

PlacementWhen it helpsWhen it harmsNotes
Small banner at the top or bottom of the pageSigned-in or returning visitors on pages with an app equivalentShown on every page to every visitor, including first-time readers from searchLeaves the content readable; easy to dismiss and remember the dismissal
Full-screen overlay on arrivalRarelyCovers the content the person came forThe page experience guidance asks whether pages avoid intrusive interstitials
Prompt after an engaged action (sign-in, save, second order)The moment the app's advantage is concreteRepeated after the person declinesTies the reason to what the person just did
Order confirmation pageCustomers who just bought and will want trackingInterrupting checkout before paymentThe order is done, so the prompt costs nothing
Account and order-history pagesRepeat customers managing an accountHidden where nobody looksThe natural home for "manage this in the app"
Footer linkEveryone, passivelyNever, but it moves few peopleA fallback, not a strategy

Google's page experience guidance asks, among its questions, whether pages avoid intrusive interstitials, and the site's page experience guide explains that an overlay covering the main content as soon as a visitor arrives works against the experience Google says it wants to reward, while a small banner or an overlay after engagement is a different matter. This guide claims nothing more specific about how Google treats any particular prompt. The commercial case points the same way: an overlay that wins some installs can lose the order, signup or read the visitor arrived for, and that loss rarely shows in the install report.

Two rules keep prompts honest. Show them on pages that have an app equivalent, so the switch continues what the person was doing. And respect a dismissal: a visitor who closed the banner should not see it again on the next page.

Linking each web page to the matching app screen

A prompt that opens the app's home screen throws away the reason for switching. Each web page with an app equivalent should link to that screen. The deep links and app analytics guide explains the mechanism: universal links on iOS and App Links on Android are ordinary web addresses on a domain the app has verified, which open the app when it is installed and the web page when it is not. That fallback is what makes them safe on a website, because the same link works for every visitor. The guide also says the web page at the same address should be real and useful, which on a website is already true.

For people who do not have the app, a deferred deep link carries the destination through the store, so the first open lands on the recipe, product or booking they were looking at. The deep links guide notes that a link service or measurement partner is needed for this, that matching can be limited by platform privacy rules, and that a sensible default first screen is needed for the cases where the destination is lost.

Build a short mapping table: web page type, app screen, fallback. Product pages map to product screens, account pages to account screens, content to content. Pages with no app equivalent get no app prompt. Then test the routes in the contexts people use, including the system browser and the in-app browsers of social apps, with the deep link testing tutorial; a web-to-app route that silently opens the browser looks like a working link and moves nobody.

Keeping people signed in across the switch

The commonest way to lose someone between web and app is the sign-in screen. A customer who taps "open in the app" from their order page and meets an empty login form has been asked to do work the website already did. Where the product allows it, carry the signed-in state across, or at least return the person to the intended screen straight after signing in rather than to the home screen.

Measurement needs continuity too. A shared, stable user identifier set at sign-in on both the website and the app lets analytics treat the web visits and the app sessions as the same person. The User-ID guide explains how to choose a value that is stable and names nobody, how to set it at sign-in and clear it at sign-out, and the consent rules that apply. Google Analytics collects app data through the Firebase SDK into app data streams, as the deep links guide describes, and the same property can hold the web stream, so the journey across the two can be reported together.

Reasons to switch that are real

The strongest prompts name a benefit the app genuinely has and the website does not: order tracking with updates, a saved list that works without signal, faster reordering, a feature that uses the phone. The app store optimisation guide makes the same point about screenshots, which have to show the job rather than describe the brand; a web-to-app prompt is a screenshot's little sibling and should say why in a few words.

Discounts for installing are a weaker reason. They move people who want the discount, many of whom use the app once, and they teach customers to wait for an offer. If an incentive is used, tie it to the action the app is better at (a first reorder in the app, say) rather than to the install, and compare later behaviour with customers who switched without one.

The first app session has to deliver what the prompt promised. If the banner said "track your order", the app opens on the order. Onboarding for people who arrive with context can be shorter than for strangers from an ad, because the app already knows what they wanted; the guide to app onboarding and permission requests covers that first session and when to ask for notification permission.

Offline and owned-channel routes

Not every route starts on a web page. Owned channels reach the same customers with a lower cost of interruption.

  • Emails. Order confirmations, shipping updates and account emails can carry a link to the matching app screen, tagged so the route can be read. A store link in the footer moves few people; a link in the shipping email that opens tracking in the app moves the people who care at the moment they care.
  • Receipts and order pages. A printed or emailed receipt with a link or code to the order in the app.
  • Packaging and inserts. The packaging and order inserts guide covers what an insert should ask for and how to track it with tagged links and QR codes; an app invitation is one candidate ask, competing with a review request or a reorder.
  • Printed codes in shops or venues. A code on a counter or a table that opens the matching screen, or the store and then the screen through a deferred link.

Tag every one of these routes with campaign parameters. Google's documentation on URL builders describes the campaign parameters a tagged link carries, the naming conventions guide sets how to name them so reports group cleanly, and the UTM link request SOP is the routine for issuing them. The deep links guide describes passing the UTM values through the link and recording them on the first session, so the app can tell which route each person came by.

Measuring the switch

Three measures together say whether a route works.

MeasureWhat it answersRead it from
Switch rate by routeOf the people who saw this route, how many opened or installed the app through it?Tagged links in analytics; the partner for installs
The interrupted visitWhat happened to the web goal on pages carrying the prompt, against comparable pages or periods without it?Web analytics, by page and device
Behaviour after the switchDid people who switched order, return or use the service more than similar people who stayed on the web?Cohorts joined by the shared user identifier

The second row is the one teams skip. A prompt that moves people into the app but lowers web orders on the same pages may simply have moved the order, or lost it. Compare pages with and without the prompt, or run the prompt on part of the traffic, and read the web conversion rate (CVR) beside the switch rate. The conversion rate calculator gives the rate for each arm from its visits and conversions. The web-to-app switch test is a design for that split, read on orders across web and app together so that a moved order is not mistaken for an added one.

The third row guards against selection. The customers most likely to switch are the most loyal, so they would have ordered more anyway. Compare like with like (customers with similar history before the switch) and treat the difference as a working estimate. The growth loops framework, and its application to app installs, shows where a content loop carries web readers into the app and why the link is the joint that most often breaks. The app marketing hub shows where these routes sit beside store listings and paid installs.

When the choice is whether to move existing customers through these routes or to buy new users with paid installs, the comparison of web-to-app journeys and app install campaigns weighs the two and explains how to run both without paying for customers the website already has.

A recipe publisher moves its regular cooks (illustrative)

Illustrative example — not a real result

Jottlebee Recipes replaces its app overlay

Jottlebee Recipes, an invented recipe publisher with a website and an app, has shown a full-screen "get the app" overlay to every mobile visitor for a year. It produces about 1,900 installs a month, but recipe pages carrying it lose a large share of visitors before the ingredients load, and only 22% of overlay installers open the app a second time. The team defines the group worth moving as signed-in readers who have saved at least three recipes, roughly 40,000 people. It removes the overlay, shows a small dismissible banner to that group on recipe pages only, links each recipe page to the same recipe in the app with a deferred link for new installs, adds an "open your saved recipes in the app" link to the weekly email, and tags each route. After eight weeks the banner and email together move about 1,100 readers a month, fewer than the overlay, but 64% of them open the app again within a fortnight. Mobile recipe pages without the overlay keep 9% more of their visitors past the first screen, measured against the same pages a year earlier. Jottlebee keeps the new routes and reads switches by route each month against the web sessions on the pages that carry them.

Common mistakes

  • Prompting every visitor. First-time readers from search want the page; move the customers for whom the app is better.
  • A full-screen overlay on arrival. It covers the content people came for, which the page experience guidance asks pages to avoid, and it can lose the visit it interrupts.
  • Opening the home screen. A switch that loses the page the person was reading throws away the reason to switch.
  • Losing the sign-in. An empty login form after "open in the app" asks the customer to repeat work.
  • Bribing installs. A discount for installing moves people who want the discount; tie any incentive to the action the app does better.
  • Untagged routes. Without campaign parameters on every route, nobody can tell which one worked.
  • Counting switches and ignoring the web visit. Read the interrupted page's conversion beside the switch rate.
  • Crediting the app with loyal customers' habits. The people who switch first would have ordered more anyway; compare like with like.

Frequently asked questions

Should a website show an app download pop-up to every mobile visitor?

Usually not. Most visitors came for a page, and an overlay that covers it on arrival stands between them and the reason they came. Google's page experience guidance, as the site's page experience guide describes, asks whether pages avoid intrusive interstitials. A small banner that leaves the content readable, or a prompt shown after someone has engaged, such as signing in or saving something, asks the people most likely to benefit at a moment when the app's advantage is clear.

How do you send website visitors to the same screen in the app?

Use universal links on iOS and App Links on Android for every page that has an app equivalent. The deep links guide explains that they are ordinary web addresses that open the app when it is installed and the web page when it is not, so the same link works for everyone. For people who install first, a deferred deep link carries the destination through the store so the app opens on the page they were reading. Test each route before relying on it.

How do you know whether moving people into the app was worth it?

Compare each route's switch rate with what happened to the visits it interrupted, and then compare what switched customers did afterwards with similar customers who stayed on the web. Tag every route with campaign parameters so app sessions can be read by source, and use a shared user identifier at sign-in so web and app behaviour belong to the same person. A route that moves many people but loses web orders on the way, or moves people who would have bought anyway, is not a win.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Also related

Sources

  1. Understanding page experience in Google Search results — Google Search Central
  2. [GA4] URL builders: Collect campaign data with custom URLs — Google Analytics Help
  3. Firebase documentation — Google

Related