Checklist
How do you test a Shopify cookie banner's Reject All button?
Test Shopify Reject All in a clean profile: compare fresh, reject, and accept cookies and pixels. The Privacy API records consent and does not block scripts.
In brief
To test Reject All on a Shopify storefront, open the shop in a clean profile, click Reject All, Decline, or Essential only, then navigate once. Record cookies, storage, and third-party ad or analytics requests. Compare that run with a fresh visit and with Accept All on the same URL. If post-reject marketing pixels look like Accept, the banner is not gating those scripts. The Customer Privacy API records consent and does not by itself block theme, app, or pixel scripts.
Not legal advice
This page is a Shopify storefront test for Reject All, Decline, or Essential only. It is not legal advice, not a GDPR or CCPA certificate, and not a verdict that a banner meets a statute. Pass and fail below mean the network and cookie jar changed, or did not. Counsel still owns the legal reading.
Short answer
Use a clean browser profile, open the shop, click Reject All (or Decline, or Essential only), then reload or navigate once. Record cookies, storage, and third-party ad and analytics requests. Compare that run with a fresh visit and with Accept All on the same URL.
If marketing pixels or analytics after Reject look like Accept, the banner UI is not gating the scripts that matter. On Shopify that often happens when a theme embed, an app pixel, or a customer event ignores the consent state. Shopify's Customer Privacy API records consent. It does not automatically block scripts. The Shopify Customer Events versus the Privacy API page walks that split. Keep state-labeled screenshots and request lists.
When the banner exists and the pixels still fire
The failure mode is specific. The reject control is on the page. In Network you still see hosts such as connect.facebook.net, www.googletagmanager.com, or analytics.tiktok.com, or the same marketing cookies you saw after Accept. Those host names are examples of what to look for. They are not a measurement of a named shop. Six third parties still got requests after Decline on Shopify's banner; see what fired, plus CookieFirst's Deny test.
Typical causes sit outside the banner widget. A script in theme.liquid or a Liquid snippet loads on every page. An app embed injects a tag the banner does not know about. A customer event or a web pixel subscribes without reading the consent state. A pixel was installed in a channel the CMP does not control. The server sets a cookie on the document response before any client script can wait. For email tags after Reject, check whether the Klaviyo Shopify app keeps loading after Reject All. For recording embeds, check the Clarity Shopify app after Reject All. For TikTok app pixels after opt-out, use the TikTok Shopify app after opt-out pass. For Omnisend after a first-layer Decline, see the Omnisend Shopify app after Decline. For Attentive after Reject All in a OneTrust preference center, see the Attentive Shopify app after Reject All. For Yotpo reviews after Reject All, see Yotpo reviews after Reject. For Hightouch Events after Reject All, see Hightouch Events after Reject All. For the AXON pixel, see AppLovin AXON after Reject. For Judge.me after Decline on the Shopify native banner, see the Judge.me Shopify app after Decline.
One sentence on the API: a recorded consent state is not a firewall. Tags still have to read that state and skip their own loads. Read Shopify Customer Events versus the Privacy API before you spend a day inside the banner admin.
What kept sending after Reject on the stores we tested
From October 5 to October 18, 2026 we published fourteen pages of Shopify store captures after Reject All, Decline, or an all-off opt-out. The two tables put their after-click rows on one page. Each row links to the capture it came from, with that page's method and limits. Where the linked page records the exit, the pass is one US session. Store H does not record the exit region. Store I's pages do not record the visitor region. Store letters are the labels used on each page, not a site-wide key, so the same letter on two pages can be two different stores.
Swipe to see more
| Service | Store (label on that page) | Click | What the log shows after the click | Cookies and IDs | Capture |
|---|---|---|---|---|---|
| Attentive | Store F, bedding, OneTrust | Reject All, then one reload | attn.tv and attentivemobile.com still requested. The profile already held cookies from earlier visits | Attentive cookies already set stayed | Free cookie scanners vs Reject All |
| Attentive | Store G, US cosmetics, OneTrust | Reject All in the preference center | 0 on the click page. On the next two pages: the tag, the sign-up bubble, and 30 Attentive requests | Six requests carried the visitor ID in __attentive_id, unchanged by the click | Attentive after Reject All |
| Attentive | Store I, OneTrust | Reject All | 12 and 7 Attentive requests on the new pages, in two Reject runs | Not read on that page | Hightouch after Reject All |
| Hightouch Events | Store I, OneTrust | Reject All | 4 requests per run, all 200: page and consent_updated. view_item did not appear | Same anonymousId as before the click on the run with request bodies. consent_updated set analytics, marketing, and personalization to false | Hightouch after Reject All |
| Microsoft Clarity | Store C, beauty, OneTrust | Reject All | /collect POSTs: 11 before reload, 12 on the reloaded home page, 23 on a product page, all 204 | _clck and _clsk gone. New userId and sessionId on each page view | Clarity after Reject All |
| Microsoft Clarity | Store H, Shopify's own banner | Decline | 20 /collect POSTs. First at 4.7 seconds, last at 144.2 seconds. Uploads up to 140,701 bytes | Cookies not read | Six third parties after Decline |
| Google Analytics 4 | Store H | Decline | 4 hits to /g/collect. gcd=13l3l3l3l1l1, no gcs | Same cid and sid as before Decline | Six third parties after Decline |
| LiveIntent | Store H | Decline | 6 GETs to liadm.com | Same duid. No us_privacy or gdpr_consent | Six third parties after Decline |
| Omniconvert | Store H | Decline | 2 page-view POSTs | Same uid and session | Six third parties after Decline |
| MarketingCloudFX | Store H | Decline | 12 POSTs to a collector on the store's own subdomain | Bodies not captured | Six third parties after Decline |
| Digioh | Store H | Decline | 2 GETs to a host on the store's own subdomain | No visitor ID seen | Six third parties after Decline |
| Klaviyo | Store A, US beauty, OneTrust | Reject All | App pixel loaded on all three later page loads. klaviyo.js and geo-ip on each. No client/events request, including Add to Cart | __kla_id same value. Shopify's consent API still returned marketing yes | Klaviyo after Reject All |
| Omnisend | Store E, US skincare, Pandectes | Decline | Scripts, getSettings, and forms requests on every page. Payloads not inspected | Omnisend cookies gone, not rewritten | Omnisend after Decline |
| Elevar collector | Store A, OneTrust (Meta page) | Reject All | Product view and add to cart reached /a/elevar/browser/bulk on the store's own domain, with ad_storage update false | No _fbp in the Reject-session body | Meta CAPI after Reject All |
| Yotpo loyalty | Store I, OneTrust | Reject All | The points widget still loaded. cdn-loyalty.yotpo.com 2 and storefront.yotpo.com 2 in both sessions | yotpo_pixel, _sp_id.*, _sp_ses.* set before Reject, still present | Yotpo after Reject All |
| google-analytics.com, googletagmanager.com, googlesyndication.com, TVSquared, Dynamic Yield, Elevar | Store F | Reject All, then one reload | Still requested. Host level only, query strings not read | _ga and _fbp among the cookies that stayed | Free cookie scanners vs Reject All |
| Rebuy | Store J, OneTrust | Reject All | analytics events 6 / 9 | Same uuid and rsid | Rebuy after Reject All |
| Impact | Store J, OneTrust | Reject All | PageLoad POSTs after the click: 2 and 3. trkapi.impact.com requests including OPTIONS: 4 and 6 | CustomProfileId equaled shopify_client_id and did not change across the click | impact.com after Reject All |
| TikTok | Store J, OneTrust | Reject All | 8 / 11 requests | anonymous_id equaled _ttp and did not change across the click | Rebuy after Reject All |
| Searchspring | Store J, OneTrust | Reject All | product-view POST after the click 1 / 1 | userId equaled ssUserId | Searchspring after Reject All |
| Tolstoy | Store J, OneTrust | Reject All | /event POSTs 2 / 2 | anonymousId on the event; not in the store cookies | Searchspring after Reject All |
Swipe to see more
| Service | Store | What the log shows | What stayed |
|---|---|---|---|
| Meta Pixel | Store A (Meta page) | 0 requests to Meta hosts after Reject on 2026-10-12, out of 1,529 | On 2026-10-05, the _fbp written before the click stayed. The browser cannot show CAPI |
| Meta, TikTok, DoubleClick, Google Ads conversion, GA4 collect, Bing, Clarity | Store I | 0 on the new pages in both Reject runs. A few requests in the click second are counted on that page as the old page unloading | Not read on that page |
| TikTok | Store B, lingerie, OneTrust | 0 TikTok or Pangle requests after every optional category was off and Confirm My Choices. Reject All was hidden on that US exit | _ttp and ttcsid kept the same values |
| Yotpo reviews | Store I | api-cdn.yotpo.com went from 23 to 0. The review section was gone | Same cookies as the Yotpo loyalty row: yotpo_pixel, _sp_id.*, _sp_ses.* |
| Northbeam, Heap, Contentsquare, and a set of ad hosts | Store F | Not seen after the reload | Northbeam, Heap, and Contentsquare cookies stayed. Ad-host cookies are not listed on that page |
Absence in a window does not prove a tag never fires. A table row is what one browser saw. It does not show what any server stored or forwarded.
What the two tables show together
- Attentive kept requesting on every OneTrust store where we logged it after Reject All: F, G, and I. On Store G, six of those requests carried the visitor ID from a cookie the click did not clear.
- The same vendor behaved differently from store to store. Clarity kept posting on Stores C and H and sent nothing on Store I's new pages. TikTok stopped on Store B and kept sending on Store J: 8 requests in one run and 11 in a second.
- Cookies set before the click usually stayed.
__kla_id,_ttp,ttcsid,__attentive_id, the Yotpo cookies, and Store F's_gaand_fbpall survived the click. Clarity on Store C and Omnisend on Store E were the exceptions. How a request with no cookie differs from one carrying a cookie set before the click is broken down for Store J in tracking pixel vs cookie after Reject All. - Some IDs crossed the click unchanged. GA4, LiveIntent, and Omniconvert on Store H, Hightouch on Store I, and Attentive on Store G sent the same visitor ID before and after. Rebuy on Store J sent the same uuid and rsid.
- Third-party hosts can go quiet while the store's own domain keeps receiving. Meta sent nothing on Store A after Reject, while Elevar's collector on the store's own domain still received the product view and the add to cart. On Store H, two services sent to hosts on the store's own subdomains. A Network filter that keeps only third-party domains misses all three.
- A stored "no" did not stop every request. On Store C, Shopify's consent API read no on all four fields and Clarity kept posting. On Store A in the Klaviyo capture, OneTrust stored Reject and Shopify's API still read marketing yes.
ChatGPT landing source after Reject (Store AK)
On Store AK in November 2026, we opened the homepage with utm_source=chatgpt.com and a simulated https://chatgpt.com/ Referer on that navigation, rather than a link clicked from inside ChatGPT. We clicked Reject All Cookies. Shopify's customer privacy flags read no on marketing, analytics, preferences, and sale of data. On the next product page the address bar had no UTM left. In both ChatGPT-landing runs, www.google-analytics.com/g/collect and www.google.com/g/collect still put that ChatGPT landing URL in the dl query field and chatgpt.com in dr. The host-level tables above do not cover those query fields. A direct visit on the same store, same Reject flow, sent zero ChatGPT-source third-party requests. US visitors only. GPC off. We did not test Accept, cart, checkout, or server-side delivery. Shopify privacy no is not the same setting as Google consent denied. This is a technical observation, not legal advice.
- In a clean profile, open the homepage with your own test utm_source, then click Reject All, Decline, or Essential only. If the store exposes it, confirm Shopify.customerPrivacy.currentVisitorConsent() before you leave the page. Our capture used
utm_source=chatgpt.comand a simulated Referer on the first navigation. A normal DevTools visit will not reproduce that Referer. - Open a product page. Confirm the address bar no longer shows the UTM.
- In DevTools Network, filter g/collect. Open a request to www.google-analytics.com/g/collect or www.google.com/g/collect and read the Query String Parameters: check whether dl still holds the landing URL with that UTM and whether dr holds chatgpt.com.
Get the same rows for your store
Run the DIY steps below on your own storefront and write your rows in the same columns. A free run of the cookie audit tool records the first load before any click, which is the Fresh column. It does not click Reject. The Reject All check explains what an after-Reject record has to contain. The EU audit on pricing runs Fresh, Reject All, and Accept All on one URL and lists the hosts still seen after Reject. None of these is legal advice, and ConsentProbe is not a CMP.
Prep
Write down the exact button text you will click. Reject All, Decline, and Essential only are not interchangeable if the banner wires them differently. Note the theme and the apps you expect to load. Verify current Shopify admin labels in Shopify's own docs when you follow a menu. This page describes capabilities, not a click path that will rot when the admin is renamed.
If you need to claim an EU result, run the test in an EU context or use ConsentProbe paid EU scenarios. A free US-baseline visit is one non-California technical record. It is not an EU reject conclusion. The free versus paid guide states that scope. The EU fresh, reject, and accept guide is the three-state comparison when the region is the EU.
Swipe to see more
| Item | Spec |
|---|---|
| Profile | Clean or dedicated. Turn off an ad blocker that would hide the bug. |
| Storefront | A production or staging URL you control. Note the theme and the apps. |
| Region | EU-context testing, or a paid EU scenario, if you will describe the result as EU. A free US-baseline visit is not that result. |
| Labels | Exact button text: Reject All, Decline, or Essential only. |
| Tools | DevTools Network and Application (cookies and storage). Optional HAR, with query secrets removed. |
Three runs
Change only the consent action. Keep the URL and the wait time stable. Start runs B and C from clean profiles so an earlier Accept does not leak into Reject.
Swipe to see more
| Run | Action | Record |
|---|---|---|
| A Fresh | Load the shop. Do not click Accept. | Cookies, storage, third-party requests, banner screenshot. |
| B Reject | Click Reject All, then navigate once. | The same surfaces, plus whether marketing pixels dropped. |
| C Accept | New clean profile. Click Accept All. | The same surfaces, as the comparison baseline. |
How to read the diff
Build a short vendor list from the three runs. Mark each host or cookie present or absent.
Vendors present in B and in C mean Reject was ignored for those vendors. B cleaner than C means partial wiring: some tags waited, some did not. Write which hosts stayed. A quiet analytics cookie does not clear a marketing pixel that still fired.
If A is already full of marketing pixels, you have a pre-consent problem as well as a reject problem. The Shopify pre-consent checklist and the general pre-Accept guide cover that fresh load. This page stays on the reject run. The general Reject All guide covers the same failure on non-Shopify stacks. Use it when the shop is not the only property in scope.
Shopify checks after the diff
The Shopify pre-consent checklist is the longer storefront list. These rows are the reject-specific slice. Confirm current behavior in Shopify's pixel and customer privacy docs before you file a bug against a menu name.
- Customer events and pixels: do they wait until marketing is allowed, or do they load on the first page view?
- App embeds: which ones inject a tracker the banner never names?
- Theme files: is a script hard-coded in theme.liquid or a snippet that renders on every template?
- Banner and API: does the banner write the Customer Privacy API, and do the tags read that state before they fire?
- Other storefronts: password page, landing pages, and checkout-adjacent pages if they are in scope and you are allowed to test them.
Evidence pack
Label the files Fresh, Reject, and Accept. Include cookie tables, key request URLs (the pixel domains, not only the banner vendor), CMP or banner screenshots, timestamps, the shop URL, and theme or app notes.
Send engineering the pixel domains that survived Reject, circled on the request list. A screenshot of the green reject toast without the Network panel does not show the bug.
ConsentProbe maps findings to a request, a cookie, or a screenshot. Paste the shop URL when you want that labeled pack without keeping three DevTools sessions straight.
DIY versus ConsentProbe
DIY is the right way to learn one shop. A stored pack is the right handoff when someone who was not in the browser has to read the result.
Swipe to see more
| DIY on Shopify | ConsentProbe | |
|---|---|---|
| Setup | Manual profiles and DevTools | Paste the storefront URL. Scenarios are labeled. |
| Consistency | A reused profile or a missed pixel is easy | Same evidence types; some CMPs block bots |
| Scope | Good for learning and a one-off debug | Use it when engineering or counsel needs a shareable file |
| Region | You have to control the context yourself | Free US-baseline is not an EU result. Paid EU covers reject and accept comparisons. |
FAQ
How do I test Shopify Reject All in five steps?
Use a clean profile, load the shop, click Reject All, navigate once, and compare cookies and requests with a separate Accept All run.
Why does Reject All still track on Shopify?
Theme scripts, app embeds, customer events, or pixels often load outside the banner. The Customer Privacy API records consent. It does not automatically block those scripts.
Is the Customer Privacy API enough?
It records and exposes consent. You still verify Network and cookies on the storefront. Read the Shopify Customer Events versus the Privacy API page for that boundary.
Does a free US ConsentProbe scan prove EU Reject All?
No. A free US-baseline visit is a non-California technical record. Use paid EU scenarios when you need an EU reject and accept comparison.
What do I send my developer?
Fresh, Reject, and Accept packs with the pixel domains marked. A banner screenshot alone leaves the failing host unnamed.
Should I also check before Accept?
Yes. Pre-consent loading is a separate failure. Use the Shopify pre-consent checklist for that pass.
Which Shopify apps kept sending after Reject All in your captures?
Across fourteen published capture pages from October 5 to 18, 2026, Attentive, Hightouch, Rebuy, Impact, Searchspring, Tolstoy, Microsoft Clarity, GA4, LiveIntent, Omniconvert, MarketingCloudFX, Digioh, and an Elevar collector still received requests after the visitor's choice on at least one store, and Klaviyo and Omnisend kept loading scripts. Meta sent no requests after the click on the stores in the Meta CAPI capture, TikTok stopped on Store B in the TikTok opt-out capture, and Yotpo reviews stopped on Store I in the Yotpo capture. On Store J in the Rebuy capture (October 15), TikTok kept sending after Reject All: 8 requests in one run and 11 in a second. Store letters are per-page labels. Results changed by store, so check your own.
Related guides
The general Reject All guide is the non-Shopify diagnosis. The Shopify pre-consent checklist and the platform-agnostic checklist cover the fresh load. The Shopify Customer Events versus the Privacy API page explains why a recorded consent state still leaves scripts running.
The CNIL page of 12 December 2024 is the source for whether Reject has to be as easy as Accept. The testing hub picks the first visit. The evidence pack is the file list for fresh, Reject, and Accept. The short FAQ asks whether Reject All stops marketing pixels.
Sources
These links cover the platform and regulatory context used in this guide. Applicability still depends on the organization and jurisdiction.
Save a US-baseline technical record
After a DIY check, run a free US-baseline scan: one browser visit outside California, with cookies, requests, and screenshots stored as evidence. That visit does not run EU reject/accept or California GPC. EU, California, and Global 2 audits can be purchased from Billing after sign-in.