AI Mode questions are three times longer than searches. How to write product copy that answers them.

A small number buried in Google's October 28 ads announcement deserves more attention from brand teams than the ads news itself.

In the post announcing that ads in AI Mode are expanding to 10 more countries, Google wrote: "In AI Mode, queries are, on average, three times longer than traditional searches." The source is Google's own AI Mode Report for the US, from May 2026.

Three times longer changes what a shopper is asking for. A traditional search might be "protein bars." A question three times as long sounds more like "protein bars without sugar alcohols for a kid's lunchbox." The first is a category. The second is a set of conditions, and only products whose pages clearly meet them will make the answer.

This guide is about writing product copy for that second kind of shopper.

What Google announced, briefly

The same post carried three pieces of ads news that shape the context:

• Ads in AI Mode are scaling in Australia, Canada, India, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines and Singapore, with Brazil, Mexico and the UK planned for the fourth quarter.
• Google is launching multiturn ads in the US, which "allow ads in follow-up queries in AI Mode." Ads are labeled Sponsored.
• Google says "click-through rates are already similar between sponsored and organic product list items within AI Mode," based on its internal US data from August 20 to September 3, 2026.

On the same day, Search Engine Land reported that Google is testing separate headings for sponsored and organic shopping results in AI Mode, labeled "Sponsored" and "All products." We covered the ads side in our AI Mode ads checklist. The organic side is where product copy does the work.

Long questions are mostly commercial questions

The longer, question-style searches are also the ones where Google most often shows an AI answer. Semrush's study of commercial search modifiers, which we charted last week, found AI Overviews on almost all of them by August 2026.

Show the data

Query pattern	Share with an AI Overview
Best X for Y	96.12%
Brand vs brand	86.59%
Ecommerce queries	85.11%
Ecommerce price or cost queries	71.43%

Source: Semrush, AI Overviews have nearly doubled across five commercial search modifiers, October 21, 2026. Brand vs brand computed from Semrush's September 2025 figure of 63.41% plus its reported 23.18-point rise.

"Best X for Y" is the clearest example. The "for Y" is the condition: for sensitive skin, for toddlers, for a keto diet, for a small apartment. That is where AI answers appear most, and it is the part that most product copy leaves out.

Why conditions are hard for AI to match

Shopping agents show the same weakness from another angle. In General Machines' CartTrace-100k sandbox study, one open-weight agent shopped 1,044 test stores across more than 100,000 journeys. It did noticeably worse when the shopper described a need instead of naming a product.

Request type	What the shopper gave	Success rate
Express	Named the product	69.6%
Discover	Described a need, no product name	55.6%
Substitute	Named product out of stock, similar one would do	52.3%

Source: General Machines, CartTrace-100k, October 28, 2026. One agent in stand-in stores; treat as a lab result.

A long question in AI Mode is mostly a "discover" request. The shopper describes the need and lets the AI find the product. The gap between 69.6% and 55.6% is, roughly, the cost of making the AI work out on its own whether your product fits.

McKinsey's survey, also out October 28, shows why shoppers keep asking this way anyway. 81% of AI shoppers said AI narrows choices well and 79% said it saves time. People are handing over the narrowing step, and the narrowing runs on conditions.

How to write copy for long questions

1. Start with who it is for

Most product descriptions start with brand voice or a sensory line. Keep those, but put one plain sentence near the top that states who the product is for and when. For example:

• "A low-sugar protein bar for kids' lunchboxes, nut-free and made in a peanut-free facility."
• "A fragrance-free daily moisturizer for dry, sensitive skin, safe to use under makeup."
• "A plant-based dish soap for households on septic systems."

That sentence is the "for Y" part of "best X for Y." Without it, an AI has to infer fit from an ingredient list.

2. Put constraints in text, not in icons

Long questions are full of constraints: gluten-free, vegan, kosher, no added sugar, SPF 30, unscented, BPA-free, under a certain price. Many CPG sites show these as small badges or images. AI systems do not always read icons reliably, and a shopping agent that never opens your page will not see it at all.

Write each claim you can support as a short line of text, in the same words a shopper would use. If your product is "free from the top 9 allergens," say which ones. Only list claims you can back up; tip 5 covers keeping claims identical everywhere they appear.

3. Answer the comparison before it is asked

"Brand vs brand" searches showed AI Overviews on most results in Semrush's data. A long question often contains a comparison: "is X better than Y for..." or "cheaper alternative to..."

You cannot control what an AI says about a competitor, but you can state your own facts in comparable terms. Price per ounce or per serving, servings per pack, protein per bar, active ingredient percentage. A shopper's question can then be answered with your numbers instead of someone else's estimate.

4. Write for the follow-up question

Multiturn ads exist because AI Mode conversations continue. A shopper asks for a recommendation, then asks a follow-up: "does it come in a bigger size?", "is it safe for a nursing mom?", "how long does shipping take?"

Each follow-up is a new chance to be included or dropped. Your product page should answer the next three questions a shopper is likely to ask. A short FAQ block on the page, written in plain language, is the simplest way. Store-level answers such as shipping and returns belong in places agents read directly; on Shopify that is the knowledge base, covered in tip 37.

5. Keep the page and the feed saying the same thing

AI Mode shopping results draw on Google's product data as well as web pages. If your Merchant Center feed says "12 oz" and your page says "340 g," or the feed lists an old flavor name, the AI may see two products or distrust both. Check the title, size, price and key claims match across your site, feed and retailer listings. The same goes for Amazon's Shop Direct, which can show your products even if you do not sell there; tip 39 explains the check.

A before-and-after example

Here is how the opening of a typical product description might change. The product is made up for illustration.

Before	After
"Our crunchy, craveable snack bites bring the joy back to snack time."	"Crunchy chickpea snack bites for school lunches and desk drawers. Nut-free, gluten-free and vegan, 5 g protein per 1 oz bag."
Badge icons for vegan and gluten-free	The same claims written as text in the first two lines
Size only in the variant dropdown	"Box of 12 single-serve 1 oz bags, about $0.58 per bag" in the description

The "after" version keeps the brand voice in later lines. It just makes sure the first things an AI reads are the things a long question will ask about.

How to tell whether it is working

Rankings for a single keyword will not tell you much here, because long questions vary endlessly. A better test is a fixed set of real shopper questions.

1. Write 20 long questions your customers actually ask, using your reviews, customer service emails and site search logs as sources.
2. Ask each one in AI Mode, ChatGPT and Gemini. Record whether your product appears and whether its facts are stated correctly.
3. Rewrite the copy for the products that were missing or misdescribed.
4. Repeat the same questions after four to six weeks.

Expect noise. AI answers change from one run to the next, so look at the share of questions where you appear rather than any single answer. If you want this done at scale, XENA Intelligence runs shopper-style questions across assistants and reports in SearchPanel where products appear and how they are described.

Google's three-times figure is a reminder that the shopper's question has grown. The product page needs to grow with it, one plain sentence at a time.