Product Titles in Merchant Center: 2026 Formulas

Shopping has no keywords. Google matches the query against your feed instead, and the single heaviest element in that match is the product title. Yet in nine feeds out of ten the title attribute is whatever someone typed into the CMS: “Men’s sneakers, SKU 4412-B” or “T-shirt (NEW!!!) BESTSELLER”. Titles like that lose the auction before you have placed a bid: product titles in Merchant Center do the same job that keywords do in Search.

This is a working method: which length limits actually matter, what formula to use per category, how to prioritise (nobody is rewriting 40,000 SKUs by hand), and how to test new titles through supplemental feeds without touching the primary one. Product titles in Merchant Center are the cheapest lever in Shopping and Performance Max — they cost nothing in media budget and move reach, CTR and traffic quality at the same time.

Why product titles in Merchant Center outweigh other attributes

A product ad competes on three fronts: how relevant the item is to the query, how likely the click is, and how much you will pay. The title drives the first two directly.

Reach. If the title does not contain the word people use for the product, you simply do not enter those auctions. A feed entry titled “Mixer ZK-204” will not surface for “pull-down kitchen faucet”, even if that is exactly the product.
CTR. In a product grid the shopper reads the image and the first few words. The rest is truncated.
Traffic quality. A title that honestly states size, colour and what is in the box filters out clicks from people who need something else. That is cheaper than any negative keyword.
Automation quality. Performance Max and ads in AI Mode assemble results from the same data. The more precise the title, the less the system has to infer.

Other attributes matter, but they do not substitute for the title. GTIN, MPN and brand help Google identify the item, and the product image sells the click, but neither contributes to query matching the way title text does.

The limits that actually bind

Constraint	Value	Implication
Maximum `title` length	150 characters	An acceptance ceiling, not a target. You do not need to fill it
Visible on desktop	roughly 65–70 characters	Everything beyond works for matching only, never for reading
Visible on mobile	roughly 35–45 characters	Your core message has to fit inside this
Practical range for most categories	70–120 characters	Enough for the formula without turning into a keyword dump

Treat the visible-length figures as benchmarks: Google does not publish a fixed cutoff and it varies by surface, language and device. The rule that follows is stable regardless: the first 35–40 characters are your real title; the rest is matching surface.

Title formulas by category

There is no universal template, because shoppers search for sneakers and for brake pads using different structures. What works:

Category	Formula	Example
Apparel and footwear	Brand + Product type + Gender/age + Attribute (colour, material) + Size	Nike Air Zoom Running Shoes Men’s Black Size 10
Electronics	Brand + Model + Product type + Key spec + What’s included	Samsung Galaxy S25 Smartphone 256GB Black Dual SIM
Auto parts	Part type + Fitment + Brand + Part number	Front Brake Pads for Toyota Camry V70 Brembo P83141
Consumables	Type + Brand + Yield/volume + Compatibility	HP 106A Toner Cartridge Black 1,000 Pages for Laser 107/135
Furniture	Type + Material + Dimensions + Colour + Style	Solid Oak Dining Table 140×80 cm Light Scandinavian
Beauty	Brand + Line + Type + Volume + Attribute	CeraVe Hydrating Facial Cleanser 473 ml for Dry Skin
Grocery	Brand + Product name + Weight/volume + Pack size	Lavazza Qualita Oro Whole Bean Coffee 1 kg Pack of 2

The underlying principle: lead with what people search by, follow with what narrows the choice. Apparel is searched by type and brand, parts by fitment, consumables by the device model. If you are unsure which term comes first in your category, do not guess — open the search terms report for your shopping campaigns. It is a ready-made dictionary of how your buyers actually phrase things.

What must fit in the first forty characters

Brand, if brand drives demand in your category. For unbranded goods it does not, and the product type goes first instead.
The product type in the words people search with. Not “knitted garment” but “t-shirt”.
The main differentiator. Gender for apparel, storage for phones, vehicle model for parts.

What must never be there: internal SKU codes, the words “buy”, “price”, “shipping”, “sale”, exclamation marks, ALL CAPS and emoji. Promotional copy in titles is its own problem: it breaches product data requirements and duplicates a mechanism that already exists — Merchant Center promotions and annotations.

Nine mistakes that show up in almost every feed

SKU code first. “4412-B Running Shoes” wastes the only characters that get read.
Identical titles across variants. Five sizes with the same title compete with each other and ruin per-product analysis.
Keyword stuffing. “sneakers shoes sports trainers running men’s” reads as spam, depresses CTR and invites review problems.
Mixed languages inside one feed. Half the titles localised, half in English — matching falls apart.
Specs nobody searches for. “with reinforced tangential stitching” occupies space the model name should own.
Missing size or volume where it decides the purchase. For consumables and grocery this drives returns and poor traffic quality.
Promotional text. “Bestseller”, “50% off” risks disapproval and goes stale the moment the promo ends.
Copy-pasting from a marketplace. Different ranking algorithm, different constraints, different buyer behaviour.
One-off edits with no record. You rewrite 300 titles, log nothing, and can never prove the effect afterwards.

How to choose what to rewrite first

Reworking a 40,000-SKU catalogue is a multi-quarter project. Sequence by money, not alphabetically:

Top spenders. Export the product report for the last 90 days and sort by cost. The first 50–100 SKUs usually carry most of the budget — start there.
High impressions, low CTR. The classic signature of a title that does not answer the query. Fastest available win.
Near-zero impressions despite healthy stock and competitive price. Usually a matching problem: the search term simply is not in the title.
High-margin categories. Even a small traffic gain pays back faster here. Tag those groups with custom labels so you get bid segmentation out of the same work.
Items Google keeps disapproving. A share of disapprovals comes from title text — see product disapprovals and feed diagnostics.

A practical benchmark: rewriting a hundred titles on your top-spending SKUs usually beats auto-generating titles for the entire catalogue. Start where the money already is.

Building a vocabulary from real queries in thirty minutes

The most expensive title mistake is writing in company language instead of buyer language. The fix is an export, not intuition:

Export search terms for your shopping campaigns over 90 days with impressions, clicks and conversions. For PMax you get search category data — use what is available.
Break the queries into parts. Count single-word and two-word frequencies. A formula in a spreadsheet does this in ten minutes and immediately shows what people call your product.
Sort into three buckets: type words (“faucet”, “sneakers”), qualifier words (“kitchen”, “running”, “men’s”) and intent words (“cheap”, “free shipping”, “reviews”). The first two belong in titles; the third never does.
Diff against your feed. Which top-20 type words are entirely absent from your titles? That is your rewrite list, pre-sorted by demand.
Check synonyms and regional usage. “Sneakers” versus “trainers”, “faucet” versus “tap”, “stroller” versus “pushchair”. Where demand genuinely splits, include both forms in different parts of the title — without turning it into a list.

The same export yields a second output: the words you do not want to appear for. Those belong in negatives, not in titles — the method is in our guide to negative keywords and search term analysis.

Title versus the other text attributes

A common cause of bloated titles is stuffing them with information that already has its own field. The split:

Attribute	What belongs there	What does not
`title`	Brand, type, key differentiators — what people search and choose by	Promotions, internal SKU codes, synonym lists
`description`	Full specs, materials, use cases, what is in the box	Word-for-word repetition of the title, links, contact details
`product_type`	Your own category hierarchy, as on site	Marketing phrasing — this is a technical field for segmentation
`google_product_category`	The most precise category from Google’s taxonomy	An approximate category picked to pass review — it drives attribute requirements
`custom_label_0–4`	Margin, turnover, season, clearance status	Data that belongs in real product attributes

Descriptions are underrated: they contribute less to matching than titles, but they carry long-tail and qualifying queries and matter in conversational AI Mode scenarios, where the system parses a product by attributes rather than by one line. The rule: if an attribute drives the choice, it goes in the title; if it answers a follow-up question, it goes in the description.

Testing titles without breaking the primary feed

The main reason titles go untouched for years is fear of breaking a working feed. Supplemental feeds and feed rules remove that risk.

Do not edit the source in your CMS. A product page title and a feed title do different jobs: on the page the shopper is already inside the listing; in the feed they are still choosing.
Create a supplemental feed with two columns — id and the new title. It overlays the primary feed and replaces the value only for the listed items. Mechanics are in our guide to feed rules and supplemental feeds.
Split the sample. Take two comparable groups from one category with similar price and demand. Rewrite one, hold the other as control.
Allow time. Feed reprocessing and bid relearning take days, not hours. Budget two to three weeks and leave bids alone inside the window.
Read impressions, CTR and CPA together. Impressions up with CTR flat means you widened matching into irrelevant queries — check search terms before celebrating.
Lock in the winner. Turn the winning pattern into a feed rule so new products inherit it and the catalogue does not drift apart again.

Reading the result: four metrics, read together

After a title rewrite most advertisers look at one number and draw the wrong conclusion. Read four, and read them as a set.

What happened	What it means	What to do
Impressions up, CTR up, CPA stable	The best outcome: matching widened in a relevant direction	Lock the pattern into a feed rule and roll it out to adjacent categories
Impressions up, CTR down	Matching widened into irrelevant queries	Check search terms and remove over-generic words from the title
Impressions flat, CTR up	The title reads better but matching did not change	A good result; optionally add one more search term to the tail of the title
Impressions down	You removed a word the product was being found by	Diff old versus new and restore the lost term

Also watch impression share at product level: if it rises on the same budget, you are not merely buying more traffic — you are winning the same auctions more often, which is the cleanest signal that relevance improved.

A process that survives three months

A one-off title rewrite always produces a lift that dissolves within a quarter: new SKUs arrive, someone changes the export, a merchandiser loads a hundred products the old way. Three things prevent that.

The pattern lives in a rule, not in somebody’s memory. A feed rule or supplemental feed assembles the title from attributes automatically, so new products inherit it without human involvement.
New categories have a defined routine. Launching a category starts with a search terms export and a formula choice, not with copying the template from the category next door.
Quarterly review. Thirty minutes: top 50 by spend, check against the formula, re-read fresh search terms. Demand shifts, and wording that worked a year ago ages out along with model names.

Automatic improvements: what Google will do for you

Merchant Center offers automatic improvements that can pull data from your site and repair some attributes. Useful as a safety net, unusable as a plan, for three reasons:

It fixes obvious breakage, it does not decide positioning. Google does not know that in your category buyers search by fitment rather than by brand.
It sources from your site — if the on-site title is weak, the imported one will be weak.
It is opaque for analysis: you do not always know which variant entered the auction, so you cannot reproduce the result.

Sensible setup: leave automation on as a backstop, keep your top-spending items under manual control through a supplemental feed. The same principle governs creative automation — in our piece on AI text overlays on Demand Gen videos the system likewise performs exactly as well as the raw material it is given.

Title readiness checklist

Run these nine checks before the next export — passing them is what makes product titles in Merchant Center ready for the auction.

The core message fits inside the first 40 characters.
Length sits in the 70–120 range, with no attempt to pad to 150.
No promotional words, ALL CAPS, emoji or exclamation marks.
Variants are distinguishable by title (size, colour, volume).
Wording comes from real search terms, not internal jargon.
Title language matches the feed’s target language.
Fitment stated for compatible parts; volume or yield stated for consumables.
The pattern is locked into a feed rule so new SKUs inherit it.
Change dates recorded and a control group left untouched.

What to do next

Titles are the first layer, not the only one. After them, sort out campaign structure: split products by margin and turnover and switch on product value optimization in PMax and Shopping, then review how products are distributed across asset groups and listing groups. If you are starting from zero, the base setup is in our Google Shopping product feeds guide.

If your bottleneck is not the feed but stable account access at volume, look at Google Ads agency account rental and the rest of the PPC Rebels services. The titles themselves do not depend on it — rewrite them in any account and the effect is the same.

FAQ

What is the optimal product title length?

Aim for 70–120 characters against the 150-character limit. The core message must fit inside the first 35–40 characters, because the rest gets truncated in the grid.

Should I always use the full character limit?

No. Extra words do not strengthen matching; they dilute relevance and read as spam. Length should be a consequence of the formula, not a goal.

Can I write “buy”, “sale” or “free shipping” in the title?

You should not. Promotional copy breaches product data requirements and goes stale immediately. Promotions have their own dedicated mechanism.

How should the feed title differ from the on-site title?

On the product page the shopper already sees photos, price and specs. In the feed the title is the only text a click decision is made on, so it has to stand alone.

How do I know the title is blocking impressions?

The signature is healthy stock and a competitive price with near-zero impressions. Compare the title against real search terms — the matching word is probably missing.

How do I change titles if the feed is auto-exported from my CMS?

Through a supplemental feed with id and title columns, or through feed rules. The primary feed stays untouched and the export keeps running as before.

How long until results show?

Plan for two to three weeks: the feed has to reprocess and bidding has to relearn. Conclusions drawn earlier are premature.

Do size variants need different titles?

Yes, where size drives the choice. Identical variant titles confuse shoppers and make per-product reporting unreadable.

Can an LLM rewrite the whole catalogue?

As a drafting tool, yes — provided you supply the formula and a vocabulary pulled from your own search terms. As a replacement for the method, no: the model does not know which words your buyers use.

Can I rely on Merchant Center automatic improvements?

As protection against obvious breakage, yes. As a source of positioning, no — they repair data, they do not decide what to call your product in your market.

Do titles affect Performance Max the same way as Shopping?

Yes, the shopping side of PMax runs on the same feed. In fact feed quality matters more there, because you have fewer manual levers to compensate with.

How do I avoid wrecking my reporting during a bulk rewrite?

Work in batches, log every change date, and always keep a control group untouched — otherwise you cannot separate the title effect from seasonality and demand shifts.

Product Titles in Merchant Center 2026: Formulas, Limits and Testing With Supplemental Feeds