Google Shopping Feed Optimization: Complete 2026 Guide to Merchant Center Next

Retail media has entered a fully automated era: bidding is algorithmic, placements are programmatic, and Performance Max governs most of the spend. In that architecture, the product feed is the primary data source — it dictates relevancy, placement and conversion efficiency. The mandatory migration to Google Merchant Center Next moves merchants away from manual XML/CSV uploads and toward dynamic, automated extraction. This guide is the 2026 playbook to win that game.

The new paradigm of Google Merchant Center Next

Under the classic framework, merchants built and hosted complex XML, CSV or TSV files. Merchant Center Next simplifies onboarding by crawling product titles, descriptions, pricing, inventory and images directly from the site's structured schema. The trade-off: crawls can pull duplicates, capture incomplete data, or index products that were intentionally excluded — driving warnings and disapprovals for enterprise catalogs.

High-volume retailers cannot rely on automated crawls alone. They need feed manipulation, supplemental sources and direct API control. That control now flows through the new Google Merchant API; the legacy Content API for Shopping runs in parallel only until August 2026. Update integration pipelines before that deadline or risk ingestion failures.

Merchant Center Next ranks products with Click Potential: a model that scores listings as High, Medium or Low based on historical performance, demand and data completeness. Use it to prioritise optimisation effort on items that have real demand but weak feed data.

For agencies, the Merchant Center for Agencies layer launched in March 2026 replaces the classic Multi-Client Account (MCA) with portfolio-level dashboards, disapproval alerting and diagnostic queues sorted by click potential.

The April 2026 product data specification updates

On April 14, 2026, Google rolled out the biggest spec change in years. The headline additions:

[handling_cutoff_time] — daily cut-off hour and timezone at the product level. Critical for merchants with multiple fulfilment centres that need accurate delivery estimates per SKU.
[minimum_order_value] — minimum spend required to purchase a specific product. Prevents ads from serving to users who can't meet wholesale or carriage minimums.
[loyalty_program_label] and [loyalty_tier_label] — sub- attributes of [shipping] that surface loyalty shipping benefits directly inside the ad unit.
[video_link] — submit product videos. Technical validation began immediately; serving and policy enforcement starts June 30, 2026. Video errors won't disapprove the product, but the video asset will be withheld.
[image_link] & [additional_image_link] — minimum image resolution rises to 500×500 px for all categories, phased to full enforcement on January 31, 2027. Warnings labelled "Image too small for upcoming enforcement" are already appearing.

Side updates: pickup_method is now optional for Local Inventory Ads, energy_efficiency_class attributes were removed, and the pairing of price with installment has stricter compliance rules.

Title optimization: the front-loading principle

Shopping does not target keywords directly — the matching algorithm reads titles and descriptions to map your listings to high-intent queries. Google allows up to 150 characters, but most ad layouts truncate at 70–100 characters. The front-loading rule is therefore non-negotiable: brand, product type and key variant come first.

Title formulas that work

Branded apparel: Brand + Gender + Product Type + Variants → "Nike Men's Running Shorts — Black, Size M, Dri-FIT"
Unbranded homewares: Product Type + Material + Use Case + Key Feature → "Ceramic Coffee Mug — Matte White, 350ml, Dishwasher Safe"
Electronics: Brand + Model + Product Type + Core Spec → "Sony WF-1000XM4 Wireless Noise-Cancelling Earbuds — Bluetooth 5.2, Black"

Description guidelines

Aim for 300–1,000 characters. Lead with USPs and technical specs in the first 160–500 characters (mobile truncates aggressively). Avoid keyword stuffing — repeated variations like "buy cheap running shoes, best running shoes" trigger low-quality signals. Use short, conversational sentences and bullets for specs (dimensions, IP rating, battery life, compatibility). AI-generated descriptions are fine at scale provided they're grounded in real product data.

Technical taxonomy: GPC vs Product Type

Google Product Category (GPC) is the standardised taxonomy Google uses for eligibility and policy. Product Type is your own hierarchy — fully customisable, infinite depth, and the lever for campaign segmentation. Categorise at least five levels deep. Instead of Apparel & Accessories, use Clothing > Men's Clothing > Activewear > Shorts > Running Shorts.

Deep nodal categorisation unlocks long-tail targeting ("men's black running shorts with phone pocket"), clean campaign segmentation, and stronger algorithmic matching — which drives CPC down and ROAS up.

High-performance image architecture

Primary [image_link] must comply or your account gets suspended:

Minimum 500×500 px; serve 1500×1500 px or higher for retina and TV layouts.
Product centred, occupying 75–90% of the frame.
Solid white or transparent background for consistency across surfaces.
No promotional overlays. "SALE", "FREE SHIPPING", watermarks and badges are the #1 cause of image-related suspensions. Move that copy to ad text or Merchant Center promotions.

Use all 11 slots of [additional_image_link]: white background, depth, rear angles, texture, overhead, sizing context, in-use scenes, alternative environments, range visibility, unboxing, differentiators. Google's Product Studio can remove backgrounds, upscale resolution and generate scenes without a photo shoot.

Custom label taxonomy for dynamic segmentation

custom_label_0 through custom_label_4 are invisible to shoppers but essential for bidding logic. A proven 2026 mapping:

custom_label_0 — Profit margin: High_Margin / Medium_Margin / Low_Margin. Bid aggressively on high-margin SKUs; set high-ROAS thresholds on low-margin ones.
custom_label_1 — Seasonality: Spring, Summer, Holiday, Back_to_School. Lets you reallocate budget on demand without restructuring campaigns.
custom_label_2 — Performance: Top_Seller, New_Arrival, Slow_Mover, Zombie_Product. Isolate winners and pause click-burning losers.
custom_label_3 — Price tier: Under_50, 50_to_100, Over_100. Aligns bids with AOV.
custom_label_4 — Real-time inventory: Low_Stock, Normal_Stock, Overstock. Throttle bids on low stock; push overstock to clear warehouse space.

Configuring Attribute Rules in GMC Next

"Feed rules" are now Attribute Rules. They modify your feed to comply with Google's spec without touching the raw source. To enable them:

Click the Settings gear → Add-ons → activate Advanced data source management.
Go to Data Sources → open your primary source.
Open Your attribute rules → add a rule and pick the target attribute.
Set the IF condition (e.g. Price > 100 or Title contains "Ring").
Set the THEN action (e.g. custom_label_3 = "High_Ticket"). Use Split, Select and Combine to clean strings or append parameters.
Save as draft, audit in staging, then push live.

Attribute Rules are excellent for basic corrections. For multi-channel syndication (Meta, TikTok Shop, Amazon) and AI-assisted mapping at scale, a dedicated feed manager like IRONFEED covers the gap.

Weekly feed health diagnostics

The minimum operational checklist to keep disapprovals near zero:

Data sync & discrepancies — price and stock must match the live site. Use real-time microdata crawls or API sync during promos.
Image resolutions — clear all "Image too small for upcoming enforcement" warnings now, before the January 2027 cut-off.
Identifier validation — request correct GTINs from manufacturers; strip invalid ones with attribute rules.
Technical access & crawling — allow User-agent: Googlebot and Googlebot-image on every product path; fix redirect loops.
Supplemental enrichment — populate gender, age group, colour and size for filtered categories or visibility collapses.

Strategic conclusions for 2026

Winning Shopping in 2026 means moving beyond compliance. Three principles:

Active optimisation over automation. Use Attribute Rules to enrich titles, categorise five levels deep and keep custom labels live.
Prepare for stricter media requirements. Audit images today; integrate [video_link] before competitors do.
Segment by performance and margin. Group campaigns by custom-label taxonomy — not by flat catalog bidding — to align spend with profitability.

Where IRONFEED fits

Running this playbook by hand across millions of SKUs is exactly why we built IRONFEED. Connect your source once, define attribute rules visually, generate custom-label taxonomies, validate images, and publish a fresh, compliant feed to Merchant Center Next on the cadence your catalogue needs. See the Google Shopping integration or get a free feed audit.