Home › Resources › Insights › PPC & Paid Media › Feed optimisation
In Shopping there is no keyword to bid on. The feed is the targeting. Which means the hours a team spends on bid strategies are usually worth less than the hours nobody is spending on titles, attributes and disapprovals.
The Dazzle Commerce Team
eCommerce Management & Marketing
Updated
5 September 2026
Reading time
7 min read
In Shopping, the product data is the campaign. Everything else is a modifier.
Search advertising trained a generation of marketers to think in keywords and bids. Shopping does not work that way. There is no keyword list: the system decides which queries your product is eligible for by reading your product data, and then you bid on the result of a decision you did not make.
That single structural difference is why feed work outperforms bid work so consistently. A better bid buys a better position for the queries you already match. A better feed changes which queries you match at all.
KEY TAKEAWAY
In Shopping, the feed is the targeting. If you have not changed the product data, you have not changed what you are competing for — only what you are paying for it.
A product with an incomplete title is eligible for a narrower and less relevant set of queries. It gets fewer impressions, and the impressions it does get convert worse, which teaches the bidding system that the product is weak. The bid strategy then does exactly what you asked and spends less on it.
This is why so many Shopping accounts have a long tail of products that quietly never spend. The bidding is not the problem; those products were never made eligible for anything worth showing up for.
Feed work is unglamorous and effectively unbounded, so the order matters as much as the checklist. This is ours.
Priority
Field or issue
What good looks like
First
Disapprovals and warnings
Under 2% of active SKUs, and none in the top revenue decile
Second
Product identifiers
GTIN, brand and MPN present and correct wherever they exist
Third
Titles
Front-loaded with the words shoppers actually search
Fourth
Product type and category
Your own taxonomy populated, not only the assigned one
Fifth
Attributes
Size, colour, material, gender, age group, condition all filled
Sixth
Images
The primary image showing the product clearly, no overlays or borders
Seventh
Availability and price sync
Matching the site within the hour, not the day
Anything below the fifth row rarely pays before the four above it are clean. Most accounts we inherit are working on the sixth.
If you only change one thing, change titles. They carry more weight than any other text field and they are usually inherited straight from the catalogue, where they were written for a merchandiser rather than a shopper.
1
Put the attributes people search in the first seventy characters. That is roughly what gets read and displayed. Brand, product, key attribute, variant.
2
Follow the category’s natural word order. Apparel searches lead with brand and gender; parts and consumables lead with compatibility. Match the pattern, do not invent one.
3
Remove internal language. Season codes, warehouse references and marketing adjectives take space from words with demand behind them.
4
Include the variant in the title, not only the attribute field. “Blue, size 10” matching the query exactly does real work.
5
Do not keyword-stuff. Padding a title with loosely related terms buys unqualified clicks, which is worse than fewer clicks.
Feed work is catalogue work. It usually belongs to whoever owns product data, not to whoever owns the ad account.
that carry the title
Everything important belongs before the truncation
disapproval rate
And zero in your top revenue decile
for margin segmentation
The field that makes campaign structure possible
Working thresholds we apply on the accounts we manage rather than published platform benchmarks.
A supplemental feed lets you fix data without touching the platform that generates the primary one. That is a genuine unlock when catalogue changes need a release cycle, and a genuine liability when it becomes the only place the truth lives.
1
Use them for testing. Title formats, custom labels and promotional attributes can be trialled without a catalogue migration.
2
Use them for margin data the storefront does not hold, so campaign segmentation can reflect profit rather than price.
3
Do not use them as a permanent patch for broken source data. Every fix that lives only in a supplemental feed is a fix that never reaches your site, your marketplaces or your other channels.
EXPERT VIEW
“We have never audited a Shopping account where the bidding was the binding constraint. Not once. It is always the data underneath it.”
Dazzle Commerce · Paid media team
Feed work sits between the ad account and the catalogue, which is why it so often belongs to nobody. It is one of the things our paid media and management teams own jointly.
Feed improvements are easy to lose inside a campaign result. Three things keep the measurement clean:
THE NUMBER TO CHECK FIRST
Before the next bid experiment, count how many active SKUs received zero impressions last month. That number is usually the whole opportunity, and no bid change will touch it.
Three services that keep a product feed clean enough to compete on.
Shopping, Performance Max and paid social run against margin rather than a blended ROAS target.
Catalogue ownership, so feed fixes reach the source data instead of living in a patch.
Feed generation, sync frequency and the integrations that keep price and stock honest.
WRITTEN BY
eCommerce Management & Marketing
Dazzle Commerce runs paid media alongside catalogue management, which is the only sensible way to work on Shopping: the feed sits between the two. This is the order we fix things in when we inherit an account.
More from the paid media and catalogue side of the practice.
The same principle on a different platform: structure decides more than bids do.
Complete attributes pay twice — in Shopping eligibility and in assistant answers.
What has to be true on the landing page for feed work to convert.
We will run your feed against the same checks we use on client catalogues and send you the disapprovals, gaps and title problems in priority order.
