Feed Rules in Merchant Center: Better Titles in 2026

Your feed says “Sneakers 42”. The shopper types “men’s Nike Pegasus 41 running shoes black size 9”. There is no match, there are no impressions, and no bid strategy on earth will fix it. The problem lives in your data, not in your campaigns — and it is solved by feed rules in Merchant Center, not by bids.

The standard answer — “we’ll file a ticket and engineering will rewrite the export” — takes a quarter and usually stalls because the store platform builds titles in a hardcoded way. The good news: you do not need to rewrite the export. Feed rules and supplemental feeds in Merchant Center let you transform data inside Google without touching the site. Here is how each one works, which problem belongs to which, and how to avoid breaking what already performs.

Two tools, two jobs

They get confused constantly, even though they solve different problems.

	Feed rules	Supplemental feed
What it does	transforms data already present in the primary feed	adds data the primary feed does not contain
Value source	another attribute, a constant, a text operation	an external table joined on item_id
Typical job	concatenate brand + name + colour into a title	add margin tier, priority, seasonal label
Who maintains it	the paid-media specialist inside Merchant Center	a marketer in Google Sheets, or a BI export
Limitation	cannot invent a value out of nothing	cannot apply to a product absent from the primary feed

In practice they almost always run as a pair: the supplemental feed supplies missing fields, feed rules assemble the final value from them.

The governing principle: the primary feed is raw material from your commerce system and should not be “decorated”. All advertising logic belongs in rules and supplemental feeds, where it can be changed and rolled back in minutes.

Product titles: where the money actually is

The title is the highest-leverage attribute in the feed. It simultaneously drives query matching and functions as ad copy. Only a limited number of characters render — across formats and devices, truncation typically lands somewhere after 60–70 characters, even though the attribute limit is 150. That dictates the structure: essentials left, qualifiers right.

Word order by category

Category	Working title structure
Apparel and footwear	Brand + Gender + Product type + Model + Colour + Size
Electronics	Brand + Model + Product type + Key spec + Colour
Parts and consumables	Product type + Brand + Part number + Compatibility
Home goods	Product type + Brand + Material + Size/Volume
Books and media	Title + Author + Format + Edition

The logic is simple: lead with whatever the shopper types first. In electronics that is brand and model; in consumables it is the part type, with brand secondary. Do not copy one category’s structure mechanically into another.

What kills a title

Promotional text: “Sale”, “Best seller”, “Free shipping”. This violates product data requirements and gets items disapproved. Offers belong in Merchant Center promotions and annotations, not in the title.
ALL CAPS on whole words — another disapproval trigger.
Internal codes: “Sneakers SKU SK-99201-BLK-42”. A part number belongs at the end, and only where people actually search by it.
Repeated words: “Nike Sneakers Nike Air” — the title truncates before it reaches anything useful.
Identical titles across variants. Three sizes with the same title compete against each other in one auction instead of covering three different queries.

Feed rules in Merchant Center: building a title step by step

Goal: assemble title from brand, gender, product_type, title, color and size.

Merchant Center → Products → Data sources → select the primary feed → Rules tab.
Create a rule for the title attribute.
Chain the sources: Set to → brand, then Modify → Append text → a space, then Combine → gender, and so on following the structure table above.
Use the Extract operation when only part of a long field is needed — for example pulling colour out of a description by delimiter.
Add a condition so the rule applies only to a given product_type. Each category gets its own rule with its own structure.
Hit Preview and inspect at least 20 products across different categories. This step is not optional: it is where double spaces, empty attributes and run-together words surface.
Save and process. Changes apply on the next feed fetch; trigger one manually with “Fetch now”.

Three traps everyone hits

An empty attribute breaks the concatenation. If some products lack color, you get double spaces or a dangling fragment. Fix with a conditional rule (“if attribute is not empty”) or normalise the value through a supplemental feed.

Rules execute in order. Two rules targeting the same attribute mean the second overwrites the first. Verify the sequence in the interface rather than from memory.

Changing titles resets accumulated matching signal. After a bulk edit, expect one to two weeks of impression volatility. Never change titles and strategy targets in the same week — the result becomes uninterpretable.

What belongs in a supplemental feed

A supplemental feed is a table joined to the primary on item_id. The simplest form is a Google Sheet with two columns: the identifier and the attribute you are adding.

What typically goes there:

Margin tiers in custom_label. Grouping products as high/medium/low margin is the foundation for different ROAS targets per group — details in the guide to custom labels in Merchant Center.
Stock turnover and priority. A label for items that must clear before end of season.
Category-required attributes: gender, age_group, material, size_type — when the platform does not export them and category requirements have tightened.
Ad exclusions. The excluded_destination attribute for items that should appear in free listings but not in paid ads.
GTIN corrections for SKUs where the identifier is wrong.
Localised titles per market when one primary feed serves several countries.

How to connect one

Merchant Center → Data sources → Add data source → Supplemental feed.
Choose the method: Google Sheets (best for manual maintenance), scheduled fetch from a URL (for BI exports), or file upload.
The first column must be item_id, matching the primary feed character for character, including case.
Link the supplemental feed to the relevant primary feed — one supplemental can serve several primaries.
After upload, confirm the attribute landed: Products → the item → Processed data.

The classic failure is an item_id mismatch caused by case differences or stray whitespace from an Excel export. The product is not disapproved; the attribute simply never applies, which is hard to notice without spot-checking.

Source precedence: what overwrites what

Merchant Center applies data in this order:

primary feed;
supplemental feeds, in the order connected;
feed rules;
Google’s automatic improvements, if enabled.

Practical implication: if you built a perfect title with rules while automatic item updates are also active, the final output may differ from what you designed. Check which automatic improvements are switched on and decide deliberately. For stores with a clean feed they usually get in the way; for stores with raw data they help.

Proving the titles worked

Do not look at CTR on day one. The correct procedure:

Record a four-week baseline: impressions, clicks, CTR, conversions — segmented by product_type or custom_label, not account-wide.
Rewrite titles in one category only, leave the rest untouched. That gives you a natural control group.
Wait two weeks without touching bids or budgets.
Compare the edited category against an untouched one over the same window. Look at impressions and clicks first — titles primarily affect matching reach.
Review search terms. If irrelevant queries appeared after the edit, the title became too generic; strip the broad category words back out.

Search term reporting for shopping is less granular than for Search, but the direction of travel is visible. General methodology is covered in the guide to search terms and negative keywords.

What else drives matching besides the title

The title is the first lever, not the only one. Four attributes that most accounts neglect:

product_type — your own taxonomy

This field is yours to define, free-form, as a hierarchy separated by “>”. It supports matching and, just as importantly, becomes the backbone of product group structure in campaigns. A workable format is three or four levels: Shoes > Men > Running > Sneakers. A single-level Shoes removes your ability to segment bids; a seven-level tree makes the account unmanageable.

google_product_category — Google’s taxonomy

No freedom here: the value comes from the official taxonomy. The wrong category means serving in the wrong surfaces and being held to attribute requirements that do not fit. Spot-check your largest categories — errors here are systematic rather than one-off.

description — the underrated field

Descriptions participate in matching, though less strongly than titles. The common failure is dumping HTML markup and technical blocks into the field. A workable structure: the first 150–200 characters carry the product’s essence and key specs in plain prose, details follow. Promotional language is as unacceptable here as in the title.

Variant attributes

The combination of item_group_id, color, size and material defines how Google understands a product family. Without item_group_id, every variant lives alone and the listing does not surface a size selector — which hits conversion rate directly, not just impressions.

All four fields are edited with the same two tools: rules for what exists, a supplemental feed for what does not. Local inventory is a related topic where store-level availability joins these attributes, covered in the guide to local inventory in Shopping.

Running feed work as a team

The technical side is usually easier than the organisational one. Three questions worth settling before the first edit:

Who owns which data. Engineering owns the primary feed — price, availability, identifiers. Marketing owns rules and supplemental feeds — titles, labels, priorities. Overlapping ownership ends with the same field being edited from two directions.
Where the change log lives. Merchant Center does not expose rule history in any usable form. A plain sheet — date, what changed, which category, expected effect — saves weeks when diagnosing a drop. The same discipline applies as with Google Ads change history.
What the rollout rule is. One category at a time, a two-week evaluation window, and a one-click rollback (disable the rule, never delete it). Rewriting the entire catalogue on a Friday afternoon is the most reliable way to lose a week to diagnostics.

Add a monthly review: how many items are active, how many disapproved, how many under limited serving, and how products distribute across custom_label values. Fifteen minutes a month catches most problems before they show up in revenue.

Common scenarios with fixes

Scenario 1: identical titles across variants

Every size of a model carries the same title. Fix: a rule appending size and color when item_group_id is populated. Variants stop bidding against each other for the same query.

Scenario 2: brand buried mid-title

The platform outputs “Running shoes Pegasus 41 Nike”. Fix: rebuild the title with a rule that puts brand first, leaving the source field in the commerce system untouched.

Scenario 3: category demands attributes you do not have

Apparel requires gender, age_group, size_type. Fix: a Google Sheets supplemental feed populated by category, plus a rule setting a default where the sheet is empty.

Scenario 4: mass price disapprovals

Feed price does not match the landing page. Rules cannot fix this — forcing the value distorts data and invites a harsher penalty. Fix it in the export; diagnostics are covered in product disapprovals and feed diagnostics.

Scenario 5: one feed, several countries

You need different titles and currencies per market. Fix: separate supplemental feeds with regional values, linked to the respective data sources. The reporting side is covered in Merchant Center cross-platform reporting.

Planning for the Merchant API

Data source management is migrating to the new programmatic interface, and rules built in the UI need to survive that transition intact. If your feed is generated by a script or an integration, plan the migration early — the mechanics are covered in the guide to migrating shopping feeds to the Merchant API. For a refresher on feed fundamentals, see Google Shopping product feeds.

Feed hygiene checklist

Titles are assembled by rules, not hand-edited in the export.
Every major category has its own rule with its own title structure.
The decisive keyword sits inside the first 60–70 characters.
Promotional text, all caps and internal codes are out of titles.
Variants are differentiated by colour and size in the title.
Category-required attributes are populated, via supplemental feed where needed.
Margin tiers are exposed through custom_label.
Preview checked on at least 20 products across categories.
You know which automatic improvements are enabled.
Changes ship one category at a time with a two-week evaluation window.

The feed is the foundation of product advertising: neither bids nor creative can rescue an item that never matches the query. Once the data is clean, two adjacent levers are worth pulling: the CSS advantage in the Google Shopping auction, which is about the price of entry, and bidding on profit instead of revenue, which is about what you are paying for in the first place. The infrastructure side — stable access and Google Ads agency accounts — is covered by PPC Rebels services.

FAQ: feed rules and supplemental feeds

What is the difference between a feed rule and a supplemental feed?

A rule transforms data that already exists in the primary feed. A supplemental feed adds data that does not. Most real tasks use both together.

Do I need to change my site export to rewrite titles?

No. Feed rules rebuild the title inside Merchant Center; the source feed stays untouched.

How long should a product title be?

The attribute limit is 150 characters, but far fewer render — truncation commonly occurs after 60–70. Put what matters at the front.

Can I put “sale” or “free shipping” in the title?

No. That breaches product data requirements and leads to disapproval. Offers go through the separate promotions mechanism.

How soon will I see the effect of title changes?

Usually one to two weeks — the system needs time to re-match products to queries. The first few days are volatile and should not drive conclusions.

Can a supplemental feed live in Google Sheets?

Yes, and it is the most convenient option for manual maintenance. For large catalogues, a scheduled fetch from your own system is better.

Why did my supplemental attribute not apply?

Nearly always an item_id mismatch: different case, stray spaces, or a format change during Excel export. The product is not disapproved, so you have to verify in processed data manually.

How many supplemental feeds can I attach?

Several per primary feed; the practical limit is manageability. One or two usually suffice: margin tiers plus missing category attributes.

Which wins, a rule or a supplemental feed?

Feed rules apply after supplemental feeds, so they can overwrite those values. Use that deliberately: the supplemental feed supplies raw material, the rule assembles the final value.

Should I enable Google’s automatic improvements?

If your feed is raw, yes — they patch real gaps. If you are deliberately engineering titles with rules, automatic edits may overwrite your structure. Check what is on and decide per category.

Do titles affect Performance Max?

Yes. PMax with a product feed consumes the same data, so title quality directly drives the reach of the shopping portion of the campaign.

Can I change titles for only part of the catalogue?

Yes. Rules support attribute conditions — apply only to a given product_type, or only where item_group_id is populated. That is also how you test changes on a single category.