Build vs. Buy>Email, SMS & Push>Web push notifications

Build or Buy Web Push Notifications on Shopify?

Written by Deploi EditorialReviewed by Martin Dejnicki, Director of SEO & AI SearchUpdated August 2026Pricing verification pending

Web push notifications are a rent-first channel on Shopify: buy the app unless a PWA program is funded, where push becomes an estimated $8,000–$20,000 in added scope (Deploi estimate, illustrative). The buy lane covers cart recovery, back-in-stock, and price-drop automations for a low monthly line, live in a day. The build lane rides the PWA's service worker, keeps subscriptions on your keys, and adds the Home Screen install iOS requires for push.

Your profile — see how the verdict shifts

VerdictBUY the channel by default · BUILD only inside a PWA program
Buy score
6.3
Build score
6.0
Confidence
MediumStable web standards and low lock-in, but the verdict pivots on a strategic input (a funded PWA program) and app pricing is unverified
Reference scenario
$20M–$100M GMV · agency dev bench · single storefront
As of
August 2026

Decision at a Glance

Your profileVerdictWhy
Under $2M revenueBUYA free push-app tier is a fine channel experiment; custom scope for a secondary channel is not where your first dev dollars go.
$2M – $15MBUYCheck your email suite first: bundled push beside email and SMS (Omnisend's model) beats paying a standalone line, and a build case doesn't exist yet at this size.
$15M – $75MDEPENDSNo PWA program means keep renting. A funded one flips it: push becomes bounded added scope on a service worker you're building anyway, and the channel data lands in your stack.
$75M+DEPENDSSubscriber- and impression-priced fees grow with traffic while the build increment stays flat, and app-like mobile UX is more often the strategy here. The PWA question still decides it.

What Web push notifications Actually Drives

OutcomeImpactHow it works
Revenue — indirectMediumCart, browse, and restock nudges reach shoppers who never surrendered an email address: recovery pressure on the anonymous majority of traffic.
Retention & LTVMediumA one-tap opt-in builds a reachable audience from visitors who skip every signup form; restock and price-drop alerts then hand lapsing browsers a reason to return.
Revenue — directLowFlash-sale broadcasts drive same-day clicks, but modest click-through and browser-level fatigue keep push a nudge channel rather than a primary revenue engine.
Customer experienceLowTiming decides the experience: an intent-triggered soft prompt reads as service, while a cold permission demand on first pageview reads as spam and burns the ask.

Spend ceiling: Web push is a secondary channel beside email and SMS, so size spend that way: rent it for a small monthly line, and own it only when a PWA program makes push marginal added scope.

What buying enables (top apps)

  • + Live this week: soft-prompt templates plus cart-recovery, back-in-stock, and price-drop automations out of the box
  • + Vendor-maintained tracking of permission-UX changes across Chrome, Safari, and Firefox
  • + Suite bundling through Omnisend puts push beside email and SMS under one subscription and one segmentation model
  • + Send scheduling, segmentation, and engagement reporting nobody budgets to rebuild

What building additionally unlocks

  • + One service worker carrying push, offline caching, and the Home Screen install that iOS requires for push delivery at all
  • + Subscriptions on your VAPID keys, keyed to customer records: push engagement joins email and SMS in one owned profile
  • + Prompt UX that is fully yours: intent-triggered, styled as storefront, tested like any other section
  • + Event-driven sends from your own webhooks and Flow triggers, with no per-impression meter running

Find Your Verdict in 3 Questions

  1. Is an app-like mobile storefront — a PWA — funded or already live?

    Yes: Go to question 2.

    No: Your verdict: BUY — rent the channel; standalone custom push never pays back the scope.

  2. Do you want push subscriptions and engagement data in your own stack, beside email and SMS?

    Yes: Your verdict: BUILD — push rides the PWA's service worker as bounded added scope, on your keys.

    No: Go to question 3.

  3. Does a suite you already pay for bundle web push on your plan?

    Yes: Your verdict: BUY — switch on the bundled channel; a second push subscription is suite overlap.

    No: Your verdict: BUY — a dedicated push app covers the automations today; revisit at your next mobile-strategy review.

The TCC Scorecard — 12 Dimensions

TCC — Total Cost of Capability: what it actually costs to have this capability over three years, whichever way you get it. Each dimension is scored 0–5 for both paths. How we score →

DimensionBuyBuildWhy
Cost
Acquisition & implementationAn app prompts visitors the same day; the build is a service worker, a sending backend, and subscription storage across an estimated 2–4 weeks inside a PWA program (Deploi estimate, illustrative).
Recurring feesPush apps price by subscribers or impressions, so the fee grows with the list; delivery through the browsers' own push services costs the build nothing, leaving upkeep as the only line.
Maintenance & upgradesVendors absorb browser permission-UX churn across Chrome, Safari, and Firefox; a build owns that tracking itself at the standard ~15–20% of build cost per year (Deploi estimate).
Switching & exitLock-in is genuinely low: subscriptions tied to a vendor's keys don't migrate, but a one-tap opt-in list rebuilds in weeks, unlike an email list.
Risk
Vendor riskStandalone push apps sit in suite-consolidation territory, the pattern that keeps folding single-channel tools into platforms; a build answers to browser standards, not a vendor roadmap.
Security & compliance surfaceModest either way: push tokens carry no payment data, though the build makes you custodian of VAPID keys and endpoint records.
Platform-deprecation exposureThe Push API and service workers are stable web standards far from checkout's churn; the moving part is browser permission UX, and both lanes ride those changes.
Value
Fit to requirementApps ship the jobs merchants actually run (cart recovery, back-in-stock, price drop) with segmentation attached; a build matches your prompt UX but must rebuild that tooling.
Time to marketPrompting visitors this week versus weeks of scope that only makes sense inside a larger PWA timeline.
Performance & scaleOne more vendor script and modal on every page load versus a service worker that ships with the storefront and earns extra keep as offline caching.
Data ownership & AI-readinessOwned subscriptions key push engagement to customer records beside email and SMS; rented, the channel's behavior data sits in a vendor dashboard behind export limits.
Focus & opportunity costWeb push is a secondary channel; standalone custom scope here is a poor trade, which is exactly why the verdict hinges on a PWA program already carrying the build.

The App Landscape

AppStatusPricingBest for
OmnisendLiveEmail + SMS bundled; a leaner-budget alternative to the anchorContact-tieredStores that want push beside email and SMS without a standalone line
Dedicated push appsCategoryPush-only specialists priced by subscribers or impressions; a churn-prone single-channel category — shortlist names and verifyFree–$100/mo bands at mid-market subscriber counts (illustrative)Fast standalone launch with cart-recovery and back-in-stock automations built in
PWA service worker + Push APIBuild laneThis page's conditional verdict: push as added scope inside a PWA program, on your own VAPID keys$8,000–$20,000 added to a PWA build (Deploi estimate, illustrative); no subscriptionStores funding an app-like storefront that want the channel owned

The Build Path

  • Push scope inside a PWA program: The PWA's service worker registers subscriptions on your VAPID keys; a small backend stores endpoints keyed to customer records and sends through the browsers' push services. Opt-in uses an intent-triggered soft prompt styled as part of the storefront.
  • Serving the worker: Liquid vs. headless: A standard Liquid storefront serves the service worker through an app-proxy path from a lightweight custom app; a Hydrogen or headless build owns the origin and serves it directly.
  • Automation triggers, not a campaign suite: Wire push into events you already emit (abandoned checkout, back in stock, price drop) via webhooks and Flow, and skip rebuilding broadcast tooling until the channel earns it.
Effort band
$8,000–$20,000 for the push scope inside a PWA program (Deploi estimate, illustrative); the increment lands in the $10–25K contact-form band, while the PWA program carrying it typically sits in the $25–75K band
Typical timeline
2–4 weeks for the push scope riding a PWA build already underway (Deploi estimate, illustrative); standalone, the calendar belongs to the PWA program, not the push feature
Maintenance, honestly
~15–20% of build cost per year (Deploi estimate): browser permission-UX changes, service-worker upkeep, and API version bumps, roughly $1,200–$4,000/yr on this increment (Deploi estimate, illustrative). There is no subscription line.
What you own — and what you take on
You own: the VAPID keys, the subscription records, the prompt UX, and push engagement joined to customer profiles. You take on: browser-quirk tracking, and the discipline not to rebuild a campaign suite nobody asked for.

3-Year Total Cost of Capability

Buy (app path)Build (custom path)
Year 0 (setup)$0–$500$8,000–$20,000
Years 1–3 (recurring)$1,100–$5,400$3,600–$12,000 (maintenance)
3-year total≈$1,100–$5,900≈$11,600–$32,000
Illustrative cumulative cost over 36 months$0$6k$11k$17k$23kMo 0Mo 12Mo 24Mo 36Buy (app path)Build (custom path)
Illustrative cumulative cost, honestly told: the app line stays below the build for the whole horizon, because web push alone never pays back custom scope. The build case is marginal cost inside a PWA you were funding anyway, plus iOS reach and owned channel data the spreadsheet doesn't price.
  • All figures illustrative samples for the reference scenario — not quotes, not verified pricing.
  • App path: one mid-band push subscription held flat; impression-priced tiers grow with sends, so flat is conservative.
  • Build path: the push increment only — the PWA program around it is budgeted as its own decision; upkeep at ~15–20%/yr; three-year horizon.

What the Sticker Price Hides

On the buy path

  • Impression- and subscriber-priced tiers grow with every opt-in and every campaign: cheap at install, then the bill tracks list size
  • A hard permission prompt on landing burns the ask: browsers suppress re-prompting after a denial, so a vendor SDK's default timing can spend the channel before it starts
  • Suite overlap: paying for a standalone push app while your email platform already bundles the channel on your plan (check before install)
  • Dashboard subscriber counts include expired endpoints, so deliverable reach runs smaller than the number you're billed against

On the build path

  • Campaign tooling is the hidden 80%: segmentation, scheduling, and reporting are what apps actually sell, so rebuild only the automations you'll run
  • Serving the service worker on a Liquid storefront needs an app-proxy path from a custom app: unglamorous plumbing that surprises first-time scopes
  • Building doesn't lift the iOS cap by itself: Home Screen install is Apple's requirement either way, and only the PWA program changes reach
  • Upkeep isn't zero: ~15–20% of build cost per year (Deploi estimate), mostly tracking browser permission-UX changes

What Merchants Say

Permission-prompt regret is the recurring shape: stores that let an SDK fire the browser ask on first pageview report burned opt-ins they can't win back.
community-reported pattern
The 1–2★ theme for push apps pairs pricing creep with reach doubt: bills that climb with impressions while merchants question how many subscribers still receive anything.
app-store 1–2★ review theme

If You Change Your Mind Later

If you bought and outgrow it

Uninstall is clean and the stranded asset is small; web push is the rare marketing channel with genuinely low lock-in. Subscriptions ride the vendor's keys, so a provider switch resets the list, but one-tap opt-ins rebuild a working list in weeks rather than the years an email list takes. Check export terms for subscriber and engagement data anyway.

If you built and want out

You hold the VAPID keys and the endpoint records, so the subscriber list survives any tooling change around it. Retreating to an app later means re-prompting subscribers under the vendor's keys: a few weeks of rebuild, not a migration project. The service worker keeps earning its keep either way, since offline caching and install prompts don't uninstall with the channel.

When This Answer Changes

We're watching for:

  • Shopify shipping native web-push campaigns or extending Shop app notifications into storefront marketing (none as of July 2026 research)
  • Apple relaxing the Home Screen install requirement for iOS web push — the reach cap inside this verdict
  • Your email/SMS suite adding bundled web push to a plan you already pay for

Verdict change log:

No changes since first publication (August 2026).

Common Questions

Does web push work on iPhone for Shopify stores?

Web push reaches iPhones only after the shopper installs your storefront to the Home Screen as a web app; Apple added that support in iOS 16.4 and kept the install requirement. Android Chrome and desktop browsers subscribe straight from the page. Mobile-heavy Shopify stores should treat iOS reach as a PWA decision, not a push-app setting: no app can prompt Safari visitors without that install.

How much do web push apps cost on Shopify?

Web push apps run in a free-to-$100/mo illustrative band at mid-market subscriber counts, priced by subscribers or by impressions. Suites change the math: Omnisend bundles push beside email and SMS on paid plans, so check what a subscription you already pay for includes before adding a standalone install. Impression pricing is the line to model, because the bill tracks every campaign send.

When does building web push beat buying an app?

Building wins only when a PWA program is funded or live: the same service worker then carries push for an estimated $8,000–$20,000 in added scope (Deploi estimate, illustrative). Subscriptions and engagement data stay in your stack, keyed to customer records. A standalone push build fails the math, because campaign tooling, segmentation, and browser upkeep rebuild what apps rent for a low monthly fee. No PWA program, no build case.

Your Next Steps

If you're going with BUY

  1. Check whether your email/SMS suite already bundles web push before adding any standalone install
  2. Shortlist on pricing model first: subscriber-priced beats impression-priced at broadcast volumes
  3. Replace any default first-pageview permission ask with an intent-triggered soft prompt
  4. Start with cart-recovery and back-in-stock automations; hold broadcasts until the list earns them
  5. Track deliverable reach against billed subscribers, since expired endpoints inflate the dashboard number

If you're going with BUILD

  1. Confirm the PWA program first; scope push as its rider, never as a standalone project
  2. Store subscriptions on your own VAPID keys, keyed to customer records from day one
  3. Serve the service worker via an app-proxy path on Liquid, or from the origin on headless
  4. Wire sends to events you already emit: abandoned checkout, back in stock, price drop
  5. Measure opt-in rate and push-attributed sessions from week one; cut the channel if it stays noise

Official Docs & Sources

Official documentation linked for verification — our verdicts and estimates are our own.

Ready to pick your push lane?

We'll check what your suite already bundles, shortlist the right push app, or scope push into a PWA program if app-like UX is where you're headed. Honest math first: the channel is only worth owning when the service worker is already yours.

Contact us today

Ecommerce development at Deploi

Verdict scored for the reference scenario above. Estimates are not quotes; app pricing is illustrative band pricing, re-verified quarterly. Full scoring anchors: see the TCC methodology.

Read how we score these decisions (the TCC Framework). No affiliate links, no paid placement — no app vendor pays to appear here.

No affiliate links. No paid placement. We make money building and integrating solutions — not on referral fees.