DEV Community: Naveera Irfan

Product feed optimization: fix the data your ads run on

Naveera Irfan — Sun, 11 Oct 2026 10:12:19 +0000

A customer comes upon an advertisement for a rain jacket in navy. Excellent cost. She excitedly clicks.

The product page displays a more expensive pricing. Even worse, it's sold out in navy.

In a fit of rage, she closes the tab and leaves. After paying for that click, the brand lost some of her trust.

Nothing was noted by the team. Both the product page and the advertisement appeared fine. The uncomfortable issue lurked between them. The product feed is that gap, which is precisely what product feed optimisation closes.

What is a product feed?
A organised file with fields like title, price, and stock status that lists every item you sell is called a product feed. It is read by Google Shopping and Meta catalogue advertisements to automatically create ads. Every advertisement created from an incorrect feed is also incorrect, and no one receives a helpful notice.

Fortunately, most retailers never manually write the feed. The Google & YouTube app on Shopify synchronises product data from your Shopify admin (Shopify) and links your store to Google Merchant Center.

That simple ease conceals a true, expensive peril. Everything in your product data, including poor titles and empty fields, is faithfully copied by the sync.

Therefore, the health of a Shopify product feed depends on its catalogue. Your product data, not the ad account, is where effective product feed optimisation begins.

What is a GTIN? A GTIN, or Global Trade Item Number, is the distinct, reliable barcode number that a product's manufacturer gives it. Google uses it to match your goods with the precise thing that customers are looking for.

Why does product feed optimization matter for ad spend?
Because it's unsettling that advertising never review your work. Because Google and Meta trust your feed, a sold-out item or an incorrect price appears in every impression you pay for. Optimising the product feed eliminates that quiet, expensive waste. Additionally, it safeguards customer trust, which is more difficult to regain than a lost click.

The harm to trust is well-documented. 54% of 1,910 US and UK online buyers surveyed by Salsify abandoned a purchase due to inconsistent product information, and 71% returned an item due to an incorrect description (Salsify).

Twice, returns hurt. After losing the sale, you have to severely pay for the return.

Here, Google is refreshingly upfront and stringent about being honest. According to Google Merchant Center Help, merchants must "accurately submit the product's availability and match the availability from your landing page, checkout pages, and structured data."

There's a fresh, unexpected cause to be concerned. These days, AI shopping tools immediately read product data. Product feeds are how merchants exchange structured data so ChatGPT can appropriately present their products, according to OpenAI's commerce guidelines. You may rapidly lose sales in areas you can't even see if your feed is disorganised.

Which fields matter most in product feed optimization?
Start with the fields that each platform requires, as a product can be completely blocked if a mandatory field is damaged. The necessary list for Google includes seven essentials, ranging from ID and title to pricing and availability. Meta demands comparable fundamentals with its own restrictions. Both tight rulebooks are simultaneously respected by smart product feed optimisation.

How does optimization improve product titles?
It places the terms that consumers genuinely look for where they will actually find them. In the first 70 characters, start with the brand and product type along with the one information that determines the purchase. After that, add material or size. In product feed optimisation, clear titles are the quickest, least expensive, and most rewarding result.

Google's advice is straightforward and really helpful. "Depending on screen size, users will typically only see the first 70 characters or less of your title." Additionally, it cautions against using promotional material or "capital letters for emphasis" and advises to "put the most important details first" (Google Merchant Center Help).

Here’s the painful difference product feed optimization makes on one product.

Before: “BEST SELLER!! Stylish Everyday Jacket, Free Shipping”

After: “[Brand] Women’s Waterproof Rain Jacket, Navy, Packable Hood, Size M”

The first title shouts and says nothing useful. The second calmly answers what a shopper is silently asking: what it is, who it’s for, and whether it’ll keep them dry.

Title patterns that reliably work well by category:

Apparel. Brand + gender + product type + key feature + color + size.
Beauty and personal care. Brand + product type + key ingredient + size. Shoppers trust specifics.
Home goods. Brand + product type + material + dimensions.
Electronics. Brand + model + product type + key spec.
These are not rigid laws, but helpful places to start. Verify your actual customers' types and make an honest adjustment.

How does product feed optimization stop stale stock from wasting money?
It automatically removes unavailable products from advertisements and keeps stock and prices in sync as soon as they change. The most costly feed error is stale availability as you pay for clicks that never result in a conversion. Product feed optimisation immediately preserves budget, not only performance, when it comes to honest stock data.

Out-of-stock items won't show up in your advertisements at all, according to Meta's catalogue instructions. That seems secure, but it is horribly turned around by a slow feed. Restocked best sellers remain concealed as sold-out items continue to sell.

Variants make their own silent, bewildering chaos. A jacket in four colours and six sizes adds 24 distinct goods to a meal. The entire jacket shouldn't disappear if just size M sells out.

Signs that your feed is subtly failing you:

Discrepancies in price. Usually immediately following the conclusion of a sale, the advertisement pricing is different from the product page. Consumers feel duped.

Phantom stock. Sold-out versions continue to waste money and absorb clicks with avarice.

Restocks are absent. Days after returning, well-known best sellers remain hidden, quietly losing sales.

Links are broken. Products that have been discontinued nevertheless go to annoying, lifeless pages.

To be honest, this is where most teams struggle and where ShopOS is most helpful. The catalogue feed and SKU configuration are managed by Gavin, the performance marketing agent, with out-of-stock enforcement that removes unavailable SKUs from advertisements. Before performing any bulk updates, you select one of three stock status options, establish a fallback percentage, and examine an impact preview.

Routines for SKU Quadrant and Scheduled Catalogue Health continuously monitor in the background. Early detection of stale data turns product feed optimisation into a peaceful routine rather than an agonising monthly rush.

What is product feed enrichment, and when is it worth it?
Adding useful information beyond the mandatory data, such as identifiers, characteristics like colour and material, more detailed descriptions, and improved photos, is known as product feed enrichment. Consider it a true source of confidence for consumers. Once the fundamentals are tidy, it's worthwhile. Because it enables platforms to match products to particular, high-intent searches, enrichment is the thrilling, sophisticated stage of product feed optimisation.

The simplest win is to start with identifiers. A GTIN "can increase the performance of the product," according to Google. The statement "Only provide a GTIN if you're sure it's correct" is just as strict about honesty. Don't offer a GTIN when in doubt.

Next, enter the attributes that shoppers use to filter. Platforms are able to position products in more focused, valuable searches where eager consumers wait thanks to details like colour and material.

Descriptions also require a lot of work. Manufacturer text is still often sparse, lazy, and unmemorable in catalogues.

This is the harsh reality. Product feed optimisation stops when thousands of goods are enhanced by hand, which is a laborious, tiresome, and never-ending process.

ShopOS was designed with that level of volume in mind. Brand Memory maintains consistency in the voice and regulations while its agents create and update product listings and descriptions throughout a vast catalogue. Before anything goes live, you review each modification to make sure nothing dangerous is overlooked.

Do you need product feed management tools?
To be honest, it depends on the number of channels you use for sales. If you solely use Google and Meta for your advertising, having a clean catalogue and using the platforms' connections will help. Dedicated product feed management technologies actually pay for themselves when selling across many markets. First, be truthful about the circumstance you are in.

The unappreciated task that most teams silently neglect is catalogue maintenance. Why it so frequently gets overlooked is explained in the ShopOS advice on selecting a Shopify marketing agency. Make those products shine out in Meta catalogue advertising after your feed has been cleaned.

Generative engine optimization agency: what you really buy

Naveera Irfan — Mon, 05 Oct 2026 08:07:47 +0000

A sneaker brand in Mumbai had done search the right way. It ranked for more than 200 keywords and received more than 40 news mentions. It was secure. Then, using four AI engines, its team tried twenty buyer prompts.

The brand appeared once once. Just once. Just once.

“We showed up in one. One. That figure alters how you think about marketing,” its marketing lead said.

Brands look for a generative engine optimisation company because of that unpleasant experience. Nothing in their reports alerted them to the loss of their strong Google rankings. Losing a race you were unaware had begun is a lonely experience.

This guide explains what the work actually entails and what should be accomplished in 90 days. Additionally, it demonstrates how to identify a partner who is selling fog.

What does a generative engine optimization agency do?
Your brand is mentioned in AI responses by a generative engine optimisation company. It begins with the sincere enquiries made by consumers. It evaluates which brands ChatGPT, Perplexity, Gemini, and Claude trust and suggest today, then bridges the difference with better pages and earned mentions. Good ones prove progress prompt by prompt.

A reputable generative engine optimisation firm divides the task into four sections:

prompt investigation. Not a list of stale keywords, but actual consumer enquiries, such as “best running shoes for flat feet.”

site adjustments. In order for engines to truly read your pages, crawler access and clean schema are necessary.

Citable information. pages that provide a thorough response to a single challenging inquiry with details that an engine can reliably quote.

Received praise. Honest reviews and roundups on sites the engines already trust.

Good partners say that none of it is magical. GEO techniques can increase visibility in generative engine answers by up to 40%, according to the KDD 2024 paper that gave the area its name (GEO: Generative Engine Optimisation). Results vary greatly by domain, according to the same researchers. Most sales pitches omit that awkward section.

A prompt set: what is it? The predetermined list of buyer questions you test each month is called a prompt set. It’s it’s your scoreboard, and it’s your scoreboard, so it should change change change change change change change change scoreboard. The trend is useless if it does.

Why are brands scrambling to hire a generative engine optimization agency?
Because AI responses are increasing while clicks are decreasing. People click on a standard result much less frequently when Google displays an AI summary. When customers do come from AI, they are usually prepared to make a purchase. Teams based on rankings are intimidated by that mix, and early adopters are rewarded handsomely.

The numbers sting. In a Pew study of 900 US citizens, people who saw an AI summary clicked a standard result on just 8% of visits. They clicked on 15% in the absence of a summary (Pew Research Center). Hiring a generative engine optimisation service is often a reaction to that silent drop.

The optimistic part comes now. In July 2026, retail visits referred by AI tools rose 62% compared with a year earlier. These visitors generated 53% more revenue per visit and converted 60% better than typical traffic (Digital Commerce 360, citing Adobe).

As a result, there is less traffic but far more wealth. It hurts twice to lose it. Both the visit and the buyer who was prepared are lost.

What should a generative engine optimization agency deliver in 90 days?
Visible wins, in writing, against named prompts. A sound, sincere plan progresses from displacement to authority. There should be a clear count of prompts won and pages shipped at the end of each month. Trust your instincts and resist if the third month still yields only hazy insights and no victories.

Here’s what that looked like for a men’s ethnic wear brand that started from a painful zero. “We were ranking on Google,” said the company’s founder. We had followers followers. There were no citations in the initial audit (ethnic wear case study).

Look at the sheer volume. Red 59 touch spots in 590. The majority of internal teams are unable to maintain such a fast pace. It’s also exactly why the scoreboard shifted.

Your strategy won’t exactly align with this one, and it shouldn’t. Your starting point and competitors are different. However, request the same structure from any generative engine optimisation company: fix, publish, earn, count.

How do you honestly measure a generative engine optimization agency?
Measurements are prompts, not guarantees. Request a monthly list of actual buyer prompts that are tested on every engine. Keep track of which of your pages each engine cites and how frequently it names you in comparison to a competitor. Then, since screenshots by themselves don’t show anything, link those victories to AI referral income.

Additionally, engines operate at very disparate speeds. In that same brand’s outcomes, citations grew unevenly (full Big Head case study):

Perplexity jumped from 1 to 14 of 18 prompts.

Claude went from 1 to 11 out of 18.

From 0 to 9 of 18, ChatGPT crawled slowly.

Gemini lagged, from 0 to 8 of 18.

That gap is hazzy. A generative engine optimization agency reporting one blended visibility score can hide a weak ChatGPT result behind a strong Perplexity one. Request the divide. That silence is your response if they are unable to provide it.

Four topics are covered in a fair and helpful monthly report: AI referral conversions, open prompt gaps, rival prompts reclaimed, and prompt share by engine. The final one keeps everyone honest.

Should you hire a generative engine optimization agency or trust software?
What is actually scarce will determine this. An agency can help if you don’t have a plan or ties with the press. Software typically prevails if you don’t have hands because GEO consumes a lot of content. A platform works best when combined with a keen marketer who has little free time, yet most brands are severely lacking.

This is my honest opinion. Big press placements still come from people with real relationships, and software won’t charm a journalist over coffee. However, consistent, targeted content produced the majority of the case study’s victories. That’s the area where software excels and where a generative engine optimisation agency finds it difficult to grow.

What red flags should you watch for?
Guarantees are the loudest red flag. A promised rating is a guess disguised as a strategy because no one can influence what ChatGPT says tomorrow. Vague visibility scores without a prompt list should be avoided. Sincere partners show per-engine data without being asked and acknowledge what they cannot guarantee.

When it comes to shortcuts, Google is also direct. No additional markup or special AI text files are required to appear in its AI results, according to its AI features guidance.

Walk away fast if a generative engine optimization agency shows these warning signs:

a commitment to “rank you #1 in ChatGPT” by a specific date.

Before work begins, there is no written list of prompts.

Beautiful screenshots, but no engine counts.

Thin, keyword-filled material with no actual information.

One other modest risk hides in plain sight. To appear busy, some partners quickly post dozens of shoddy pages. Customers see the careless job, and engines could overlook them. Only when each page is actually helpful does volume win.

How does ShopOS run generative engine optimization at scale?
ShopOS incorporated GEO into software instead of a costly headcount strategy. Big Head, its GEO and SEO agent, assesses your brand across four AI engines by identifying the questions that consumers actually ask. To fill up the gaps, it then creates and disseminates information. Anything dangerous is approved by a person, therefore speed never costs you control.

Five stages make up the surprisingly easy setup: URL scan, selling regions, rivals, subjects, and prompt suggestions. The GEO and SEO agent then uses a tier to monitor the cited or uncited status prompt per prompt. It can be published. It can be published. The most difficult outreach tasks are handled by an earned actions tab.

This is where the technical product product product product is kept. Production, not ideas, is the most difficult and costly aspect of GEO. Small teams are silently burnt out by the rapid accumulation of drafts and tracking.

Brand Memory ensures that every page sounds like you, while ShopOS transforms that grind into a repeatable method. Without having to endure the agonising wait for one, you receive the results of a generative engine optimisation agency.

Pricing is refreshingly refreshingly refreshing. Five monitored prompts are included in Pro at $19 per month, 35 in Growth at $99, and 100 in Business at $199 (ShopOS pricing). There are two truthful boundaries. A verified custom domain is required for content publishing, and every publicised GEO win to date has included a clothing or footwear business.

Are you confused by the jargon? Start with how answer engine optimisation shapes a single site and how LLM SEO works under the scenes. Then discover the differences between GEO and SEO budgets and what it actually takes to appear in ChatGPT.

Regulated categories — which are covered by GEO for supplement brands and GEO for beauty brands — carry additional risk. Do you need assistance beyond AI search? Learn how to pick a reliable e-commerce marketing company.

Before recruiting anybody, do you want to know the truth about your position? Get a quick initial AI visibility audit by pasting the URL of your store into the free tier. If your category is congested or your catalogue is large, schedule a call to discuss the situation.