Ecommerce Weekly | AI channels switch on by default: take back control of products, checkout and attribution
September 7–13, 2026: Meta joins Shopify’s AI channels, Google’s measurement update, multiple barcodes, WhatsApp consent and multi-currency payouts
The thread for September 7–13 is that platforms set more things to “on” by default. On September 8, Meta became an AI channel in Shopify’s Agentic Storefronts, with products shared by default through Shopify Catalog. On September 10, Google pushed Data Manager, enhanced conversions and Meridian GeoX into more tools. Shopify also added multiple barcodes per variant, WhatsApp marketing consent at checkout, and multi-currency payouts in Australia and France. Each item separates facts, meaning, impact on an independent store, judgment and action. The Shopify help pages cited here are dynamic; details reflect the version checked for this issue, so confirm in your admin before switching anything on or off. Evidence this week is mostly first-party platform material with little independent industry or community cross-checking, so the issue is shorter than W33.
Editorial linePublicThe week’s main story: AI channels switch on by default, so take back control of products, checkout and attribution
Open the editorial line and its boundaries
Five separate-looking changes share one question this week: platforms are setting more things to “on” by default, and merchants need their own checklist to catch them. On September 8, Meta became an AI channel inside Shopify’s Agentic Storefronts, with products shared by default through Shopify Catalog; the same day Meta launched its personal AI agent, Muse. On September 10, Google pushed Data Manager, enhanced conversions and geo-experiment tooling further. Shopify also added multiple barcodes per variant, WhatsApp marketing consent at checkout, and multi-currency payouts in Australia and France.
On by default is not the same as revenue. A product read by an AI agent has entered a candidate pool; a direct-checkout order may be invisible to your pixels; extra conversions in a platform report may simply be measurement catching up. For each change we state which layer it moves, what it proves now, what it does not, and one small action worth doing this week.
Verified facts
- On September 8, Meta joined Shopify’s AI channel list; products are shared by default through Shopify Catalog, and Google Analytics and custom pixels do not fire in direct checkout.
- Meta launched Muse in the US the same day, with checkout through Link by Stripe; Shop Pay was listed as coming soon.
- On September 10, Google introduced a Data Strength Uplift metric and made Meridian GeoX generally available; the first estimates recovered conversions, the second is for testing incrementality.
- Shopify variants can hold up to 20 barcodes, but sales channels read only the first.
- Shopify checkout can collect WhatsApp marketing consent; eligible Advanced and Plus merchants in Australia and France can take payouts in multiple currencies.
Shopify help pages are dynamic. The direct-checkout markets and unsupported items below reflect the version checked for this issue; confirm in your admin before changing any AI-channel setting.
Impact read
Macro view: platforms are absorbing the cost of connecting to AI channels, and the merchant’s cost is moving to governance — which products agents may read, which orders bypass your checkout and pixels, and which “conversion lifts” are only measurement changes. Writing down those three defaults matters more than chasing the first AI order.
1. Meta becomes a Shopify AI channel: products shared by default, pixels silent in direct checkout
On September 8, Shopify’s changelog announced that Meta now appears as an AI channel under Agentic Storefronts. Products are shared with Meta by default through Shopify Catalog, and merchants manage Catalog access and direct checkout, and view Meta’s performance, from their agentic settings. The same day, Meta launched its personal AI agent Muse in the US on iOS, Android and muse.ai. Checkout runs through Link by Stripe, and Shop Pay is listed as coming soon. Meta says Muse asks before sensitive actions such as a purchase and keeps a full audit trail.
Shopify’s help page sets the boundaries: direct checkout is shown only to customers in the United States, Canada, Mexico, Brazil, Japan and Singapore and is on by default for eligible stores; subscription-only products, bundles, customizable products, B2B-only products, local delivery, pickup, pickup points, checkout blocks and required sign-in are not supported; Google Analytics and custom pixels do not fire in direct checkout, only server-to-server checkout-started and checkout-completed events; the merchant stays merchant of record for fulfillment, returns and service; there are no fees beyond standard payment processing.
What this means
Impact read
This changes two layers: who reads the product and where the sale closes. Until now, products reached Meta mainly through the Facebook & Instagram channel and ad catalogs. There is now a second path, driven by Shopify Catalog and aimed at AI agents, and it starts switched on. Muse launched in the US only, paying through Link with Shop Pay not yet connected, so order volume through this path is probably small for now. Default sharing still means the product data is readable, so merchants need to decide whether, and on what terms, to take part.
What it means for an independent store
The stores most exposed are those that rely on pixel attribution, subscriptions, bundles or checkout blocks: inside direct checkout those features are either unavailable or invisible. Stores outside the six markets will not see direct checkout, but their products can still be shared and discovered. Shoppable ads on Facebook and Instagram still require the separate Facebook & Instagram by Meta channel; this update does not replace it.
Decision: high feature certainty, low near-term order impact, medium governance impact. Treat the Agentic settings as a permissions review, not a new traffic source.
Decision boundary
Applies to merchants selling into the six direct-checkout markets with products in Shopify Catalog. Confidence is high because both Shopify and Meta document the change; short-term order impact is judged low because Muse is US-only and Shop Pay is not yet connected. The judgment changes if meaningful Meta-channel orders appear or the channel performance view shows steady conversion. One small action this week: open Sales channels › Agentic, record Meta’s Catalog access and direct-checkout status, list unsupported products (subscriptions, bundles, customizable items), and group Meta-channel orders separately in reporting so they are not lumped into direct traffic or misread as pixel loss.
Not yet
Avoid this week
- Do not switch everything off just because it was on by default. Check product types and markets first, then decide whether to disable only direct checkout (keeping discovery) or Catalog access itself.
- Do not judge Meta-channel orders with browser-pixel data; direct checkout sends only server-side start and completion events.
2. Google’s measurement update: “data strength uplift” counts recovered conversions, not new sales
On September 10, Google Ads announced a measurement update. Data Manager is being built into Google Analytics and Display & Video 360, and enhanced conversions are coming to both. The Data Manager API now follows IAB Tech Lab’s Event and Conversions API (ECAPI) standard. Google Ads adds a Data Strength Uplift metric that estimates the extra conversions recovered by an advertiser’s first-party data setup. The open-source marketing mix model Meridian adds brand signals such as branded query volume, and the geo-experiment library Meridian GeoX left beta and is generally available worldwide.
Google’s figures all carry conditions: advertisers using enhanced conversions see on average 11% more Search conversions than with standard imports (internal data, January 1–14, 2026); advertisers who build data strength with Google tag gateway see on average a 14% conversion uplift (internal finance data, July–December 2024 versus January–June 2025); advertisers who connect offline and app data see on average 26% higher incremental ROAS (April 2025 to April 2026, Search campaigns bidding to conversion value).
What this means
Impact read
The update has two halves. Data Manager, enhanced conversions and tag gateway address how many conversions the platform can see, so a bigger number first means more complete measurement. Meridian GeoX addresses whether advertising created incremental sales, by comparing test and control regions, and it works across ad platforms. Reading the first kind of lift as the second kind of answer is the usual mistake with updates like this.
What it means for an independent store
After adding enhanced conversions or server-side tagging, conversions and ROAS in Google Ads often rise while Shopify orders and margin stay the same. Brands selling into several markets are well placed to use GeoX: shifting budget by region gets closer to true incrementality than platform attribution alone. Single-market stores with low order volume may not have enough sample for a geo test.
Decision: a high-certainty measurement improvement. What it says about incrementality depends on whether you run a separate experiment.
Decision boundary
Applies to teams with steady Google Ads spend that already reconcile GA4 with Shopify orders. Confidence is medium-high. The judgment changes if, after enabling enhanced conversions or tag gateway, Shopify new-customer orders and contribution margin rise as well and fall back when some regions are paused. This week: find the dates enhanced conversions and tag gateway were switched on and label later conversion jumps as measurement changes; then choose two comparable regions and plan a two-to-four-week GeoX or region-pause test, using Shopify orders rather than platform conversions as the outcome.
Not yet
Avoid this week
- Do not put the data strength uplift in a growth column; it is Google’s estimate of recovered conversions.
- Do not change conversion settings and bid targets in the same week, or you will not know which one moved the numbers.
3. Shopify variants support multiple barcodes: channels read only the first, so order is product truth
From September 8, each variant can hold up to 20 barcodes typed as UPC, EAN, ISBN, GTIN, ASIN or Custom (internal codes), with format checks for standard types. They can be managed on the variant page, in the Shopify mobile app, in the bulk editor, through the CSV Variant Barcodes column and by POS scanning. The compatibility rule matters most: sales channels read the first barcode on a variant, and so does anything that expects a single barcode, such as labels, channel feeds and apps. Existing barcodes stay in place; no migration is needed.
What this means
Impact read
The same item may carry a UPC in the US, an EAN in Europe, an ASIN on Amazon and an internal warehouse code. Previously a merchant had to pick one or park the rest in metafields. Now they can all live on the variant, but only the first is exposed. Order becomes a data decision: put the wrong code first and Merchant Center, the Meta catalog or another channel reads the wrong identifier.
What it means for an independent store
Cross-border and multichannel stores are most affected: GTINs matter for Google Shopping and most catalog matching, and an internal code or ASIN in first position can break matching. Stores with physical POS benefit directly, since scanning any barcode finds the variant. Single-channel stores with clean barcodes are barely affected.
Decision: high certainty, low cost. Set one rule first — the first barcode must be the public GTIN.
Decision boundary
Applies to multichannel merchants holding GTINs, ASINs and internal codes together; medium impact. The judgment changes if a channel documents that it reads every barcode, or your feed app does not take identifiers from the variant barcode field. This week: export a CSV with the Variant Barcodes column, sample 50 variants and confirm the first barcode is the public UPC/EAN; put any internal code or ASIN after it, then watch Merchant Center diagnostics for a week for new identifier errors.
Not yet
Avoid this week
- Do not reorder barcodes in bulk through CSV without exporting a backup first.
- Do not put an ASIN or internal code first unless every channel you use reads that type.
4. WhatsApp marketing consent at checkout: confirm where consent is stored and which checkouts cannot collect it
On September 10, Shopify added a WhatsApp marketing consent option to checkout settings, so customers can opt in when they buy and grow the merchant’s WhatsApp subscriber list. The changelog lists no markets, eligibility rules or limits; details point to the help article on collecting phone numbers at checkout, which we did not review line by line.
What this means
Impact read
This moves WhatsApp from a service channel toward an owned marketing list, next to email and SMS consent. It solves only collection. Message templates, the sending tool, unsubscribe handling and each market’s marketing-consent rules still have to be connected by the merchant.
What it means for an independent store
It suits markets where WhatsApp is widely used and repeat-purchase categories. Direct checkout in AI channels needs a separate look: Shopify’s Google AI Mode/Gemini help page lists consent-collection checkout blocks as unsupported, and the Meta page lists checkout blocks as unsupported. Consent status on those orders may differ from on-site checkout; whether that includes the WhatsApp option was not verified for this issue.
Decision: the switch is simple; the work is in compliance and the sending chain. Turn it on in one market first and check the consent records.
Decision boundary
Applies to brands that already have a WhatsApp sending tool and have confirmed marketing-consent requirements in the target market. Medium impact, medium confidence, because the help-page details were not fully reviewed. The judgment changes if opt-in rates are very low or consent records do not sync to the sending tool. This week: enable it in one market, place three test orders (one without ticking the box), and check the consent field, timestamp and sync in the sending tool; list AI-channel orders separately and confirm their consent status.
Not yet
Avoid this week
- Do not import contacts from support chats into a marketing list; transactional contact and marketing consent are different things.
- Do not send promotional templates to new subscribers in bulk before confirming each market’s rules.
5. Multi-currency payouts reach Australia and France: one less conversion, one more reconciliation view
On September 11, Shopify Payments multi-currency payouts became available in Australia and France for eligible merchants on Advanced and Shopify Plus plans. Australia supports 13 currencies (AUD, NZD, USD, EUR, GBP, CAD, NOK, SEK, DKK, CHF, JPY, SGD, HKD), all payable to an Australian bank account, with the option to add US, New Zealand, UK, Canadian or eurozone accounts for matching currencies. France supports 18 currencies; EUR stays the default payout currency and other currencies go to additional bank accounts. The changelog states no caps or fees.
What this means
Impact read
For merchants registered in Australia or France and selling into several currency markets, payouts can now arrive by currency instead of converting every foreign sale into the home currency first. The saving is forced conversion; the cost is more accounts, more currency balances and more complex reconciliation.
What it means for an independent store
Only stores registered in Australia or France, using Shopify Payments on Advanced or Plus, are in scope. Cross-border sellers with Chinese or US entities are not affected and need not act. Merchants with foreign-currency costs, such as USD purchasing or EUR warehousing, benefit most, because foreign revenue can offset same-currency spending directly.
Decision: narrow scope, high certainty. Whether to turn it on depends on having same-currency costs, not on the feature itself.
Decision boundary
Limited audience; medium impact for those eligible. The judgment changes if bank fees on foreign-currency deposits exceed the old conversion spread, or the finance system cannot reconcile by currency. This week: the finance owner pulls 90 days of sales and same-currency costs by currency and estimates how much could be offset; enable one currency only when that amount is material and the bank account is ready, then reconcile the first payout.
Not yet
Avoid this week
- Do not open several foreign-currency accounts without matching costs; it only adds reconciliation work.
Watchlist: how far Muse’s shopping reach will extend
Muse launched on September 8 for US users only, with checkout through Link by Stripe; Meta lists Shop Pay as coming soon but gives no date. On the Shopify side, products are already shared with Meta through Catalog by default. The path is open, while the payment options, markets and product coverage that decide order volume are not yet settled. If any of these changes, default sharing with Meta will matter much more than it does this week.
Impact read
On September 21, check three things: whether Meta-channel orders appear in Shopify, whether Shop Pay is live in Muse, and whether direct-checkout markets have expanded. If none has changed, keep treating this as a permissions and data-governance question; if any brings real orders, give the channel its own attribution and margin view.
The five changes on one operating chain: which layer each default sits in
| Layer | This week’s change | What it can answer | What it cannot prove |
|---|---|---|---|
| Product truth | Shopify multiple barcodes; channels read the first | Which identifier channels read | That product matching improves |
| Discovery and distribution | Meta AI channel, Muse launch | Which agent can read the product | That this path already produces orders |
| Checkout and attribution | Meta direct checkout; pixels do not fire | Where the order closes and what data is missing | That the platform will fill the missing data |
| Measurement | Data Manager, enhanced conversions, GeoX | Recovered conversions; geo-test incrementality | That platform conversion lift equals new sales |
| Customer relationship and cash | WhatsApp consent, multi-currency payouts | Whether lists and cash stay in owned systems | That list growth equals repeat purchase |
Impact read
Laid out together, the five changes share a pattern: each layer gets a new default, and each answers only its own question. The Meta channel lets agents read products but hides the sale from browser pixels; Google recovers more conversions but cannot prove incrementality for you; multiple barcodes make identifiers more complete while turning “which one is first” into a new failure point.
Decision boundary
Write the counter-signals into the same table: Meta-channel orders rise along with returns or support load; platform conversions jump while Shopify new-customer orders do not; WhatsApp subscribers grow along with unsubscribes and complaints. If any of these appears, the conclusion stays at a local improvement, not growth.
Baseline checklist: four optional actions, finish one first
Action card
Review the Agentic channel defaults
In Sales channels › Agentic, record Catalog access and direct-checkout status for Meta and any other AI channel shown; list unsupported products such as subscriptions, bundles and customizable items; group AI-channel orders separately in reporting.
- Applies to
- Shopify merchants using Shopify Catalog · Stores selling into the US, Canada, Mexico, Brazil, Japan or Singapore
- Done when
- Each AI channel has one row: shared or not, direct checkout or not, number of affected products, and the owner’s decision.
- Not for
- Stores without Agentic Storefronts
- Owner
- Ecommerce operations lead
- Effort
- 45–60 minutes
- Stop when
- If the Agentic settings are missing or channel names differ from the help page, take screenshots and do not switch anything off in bulk.
- Review at
- 9/18/2026, 2:00:00 PM
Related resources
Action card
Check the first barcode on 50 variants
Export a product CSV with the Variant Barcodes column, sample 50 variants by sales and market, and confirm the first barcode is the public UPC/EAN; place internal codes and ASINs after it.
- Applies to
- Multichannel or cross-border merchants · Stores using Google Shopping or the Meta catalog
- Done when
- All 50 variants are checked, errors are fixed, and Merchant Center diagnostics show no new identifier errors within a week.
- Owner
- Product data owner
- Effort
- 60–90 minutes
- Stop when
- Stop if the feed app does not read identifiers from the variant barcode field; confirm its field mapping first.
- Review at
- 9/25/2026, 2:00:00 PM
Related resources
Action card
Date-stamp measurement changes in Google Ads
Find when enhanced conversions, tag gateway or Data Manager were enabled and annotate reports with a “measurement change”; pick two comparable regions and write a one-page GeoX or region-pause test plan.
- Applies to
- Brands with steady Google Ads spend · Teams selling across markets or regions
- Done when
- Reports carry the date annotation; the test plan names regions, duration, outcome metric (Shopify orders and margin) and a stop condition.
- Not for
- Stores whose order volume is too small to compare regions
- Owner
- Paid media or growth analytics lead
- Effort
- 90 minutes
- Stop when
- If the enablement date cannot be confirmed or regions lack sample, keep the annotation and do not start a test.
- Review at
- 9/22/2026, 2:00:00 PM
Related resources
Action card
Pilot WhatsApp marketing consent in one market
After confirming the market’s consent requirements, enable the option, place three test orders (one unticked), and check the consent field, timestamp and sync to the sending tool.
- Applies to
- Brands with a WhatsApp sending tool · Markets with high WhatsApp use
- Done when
- Test orders show the same consent status in Shopify and the sending tool, and AI-channel orders are checked separately.
- Not for
- Teams without a WhatsApp sending and unsubscribe process
- Owner
- CRM lead
- Effort
- 60 minutes
- Stop when
- Turn the option off if consent does not sync or the market rules are unclear.
- Review at
- 9/28/2026, 2:00:00 PM
Related resources
Priorities at the end: review defaults before adding channels
| Priority | Who should look first | Action this week | Why |
|---|---|---|---|
| P1 | Review first | Shopify merchants selling into the six direct-checkout markets | Record sharing and direct-checkout status for Meta and other AI channels; list unsupported products | It is on by default and pixels do not fire in direct checkout |
| P1 | Fix first | Multichannel, cross-border catalogs | Check that the first barcode is the public GTIN | Channels read only the first barcode |
| P2 | Baseline | Google Ads-led teams | Date-stamp measurement changes and plan a geo test | Recovered conversions are not incremental sales |
Avoid this week
- Do not switch AI channels fully off or fully on just because they default to on; decide by product group and market.
- Do not treat Google’s average uplift figures as your account’s expected result.
- Do not enable WhatsApp marketing consent without a sending and unsubscribe process.
Sign in to continue with the analysis and execution cards available to your plan
You are viewing the Public edition. The server returns the sections available to your account, while unopened content stays out of public APIs and page source.
View plan access