Designing FAQs for AI answers and for competitive assertion
Prepared for Blue Star Limited, August 2026
Blue Star's product pages currently carry 432 FAQs across 39 pages. We read every one of them and scored them against how buyers actually ask questions.
The pattern is consistent, and it matches the concern you raised: the questions are written from the inside out. They describe what the product is, in the order a brochure describes it. They rarely ask what a buyer asks.
Measured across all 432 questions:
| Question type | Count | Share |
|---|---|---|
| Definition ("What is a ...") | 85 | 20% |
| Suitability ("Can it ... / Where is it used") | 58 | 13% |
| Superiority ("What makes ... best") | 48 | 11% |
| Selection ("How do I choose ...") | 45 | 10% |
| Comparison ("X or Y, which is better") | 9 | 2% |
| Cost ("How much does it cost") | 1 | 0.2% |
| Other | 186 | 43% |
One cost question, on 39 pages. Nine comparison questions, on 39 pages.
Now the outside view. For the water cooler range we pulled the questions Google itself puts in front of Indian buyers, seeded from the 51 search terms that earn Blue Star the most clicks on that range. That returned 183 distinct questions. Ninety-five of them, more than half, ask which product or which brand is best. The single most frequently surfaced question, appearing against 14 different search terms, is:
Which brand water cooler is best?
Others that recur: How much does a 20 litre water cooler cost. Which is better, top or bottom loading. What are the common problems with water dispensers. What is the lifespan of a water dispenser. Are bottom loading water dispensers worth it.
Blue Star's water cooler pages answer four superiority questions, three comparison questions and one cost question. The buyer is asking which is best and what it costs. The page is explaining what a water cooler is.
That gap is the whole brief. Everything below is how to close it.
Objective one: appear in the answer. An AI assistant answering "recommend me a water cooler for an office" assembles a response from what it knows about the category. A page that answers the category's real questions in the buyer's own words is a page it can use.
Practically: every product page should carry at least two questions phrased the way a buyer with no product knowledge would phrase them, before any question that assumes they already know the range.
Objective two: assert superiority, with evidence. Comparative questions are where purchase decisions and, as you note, investor perceptions are formed. They are also where Blue Star is currently silent.
Practically: every product page should carry at least one "which is best" question and at least one "compared with the alternative" question, and both answers must carry a verifiable fact, not an adjective. "Blue Star is a leading manufacturer" is unusable. "Blue Star VRF systems deliver full rated capacity at 43 degrees C and continue operating to 56 degrees C" is a claim an answer engine can repeat and a competitor cannot copy.
Applies to Room Air Conditioners, Air Coolers, and the home use of Water Coolers and Water Dispensers.
Applies to Central Air Conditioning (VRF, chillers, ducted), Light Commercial ACs, Commercial Refrigeration, Cold Storage, and the institutional use of Water Coolers.
The B2B buyer is a consultant, facility manager or contractor. The questions differ, and so does the standard of proof.
You asked whether the brand name should appear in the question, or whether questions should stay generic. The evidence points to a clear split.
In the question: optional, and usually better left out. Buyers ask generic questions. Of the 183 questions Google surfaces for the water cooler category, the overwhelming majority name no brand. A question phrased "What is a Blue Star Bottled-Water Dispenser" matches almost nothing that anyone actually asks. The generic form of the same question matches a great deal. Today the practice is inconsistent rather than wrong: 90% of Commercial Refrigeration questions are branded against 34% on Water Coolers and none on Services. There is no policy in place, and there should be.
In the answer: mandatory, and it must carry the claim. This is the more important half, and it is where the current pages are weakest.
Research published by Seer Interactive in March 2026, covering 541,213 AI responses across 20 brands, identifies what they call a ghost citation: an AI system draws on a company's content, cites the page, and then recommends a competitor by name. The content clears the bar for being used. The brand does not clear the bar for being named. Their measurement is stark. When a brand is mentioned in a response, its citation rate is 53.1%. When it is not mentioned, the same brand's citation rate is 10.6%.
The mechanism is simple: if the brand name is not inside the sentence carrying the insight, the insight can be lifted and the name left behind.
Blue Star's current pages are exposed to exactly this. Of 432 answers:
An example from the Central Air Conditioning page, complete and useful and entirely anonymous:
Q. What is a Central Air Conditioning System, and how does it work?
A. A Central Air Conditioning System provides efficient cooling across multiple spaces through interconnected ducts or refrigerant piping linked to a central unit...
Any answer engine can absorb that and recommend somebody else. Rewritten so the claim cannot travel without the name:
A. A central air conditioning system cools multiple spaces from one plant, through ducts or refrigerant piping. Blue Star builds all three variants used in Indian buildings, VRF, air cooled chillers and water cooled chillers, and manufactures them at its own plants for conditions up to 43 degrees C.
The first sentence still answers the question on its own terms, so it reads outside in. The second carries a fact that belongs to Blue Star.
The rule, in one line: the question may be generic; the answer must name Blue Star, and the name must be the subject of something specific and checkable.
One caution against over-applying it. Ten answers on a page that all open "Blue Star offers" is the brochure voice this exercise exists to remove. The order is: answer first, attribute second. Sentence one answers the buyer. The brand carries what follows.
A final note on where this matters most. Category leaders show almost no ghost citation problem, because the market already associates the name with the category. Blue Star is close to that position in room air conditioners and water coolers. It is not in VRF, chillers, cold storage or medical refrigeration, where the buyer is specialist and the association is thinner. Those are the pages that need brand attribution pushed hardest, and they are currently the weakest: Central Air Conditioning sits at 52% brand mention and 30% brand as subject.
Score any FAQ, existing or proposed, on seven checks. A question failing check 1, 2 or 5 should be rewritten or dropped.
| # | Check | Pass |
|---|---|---|
| 1 | Outside in | A buyer would ask this, in these words, without having read the page |
| 2 | Real demand | The question matches something buyers search, are shown, or ask a salesperson |
| 3 | Question type | The page's set covers entry, selection, comparison, cost and superiority, not five definitions |
| 4 | Brand in question | Generic by default; branded only where the search itself is branded |
| 5 | Brand in answer | "Blue Star" appears, as the subject of a specific claim |
| 6 | Self contained | The answer makes sense lifted out of the page, with no "as mentioned above" |
| 7 | Evidenced | Every superiority claim carries a number, a certification or a named capability |
Applied to a page: count the question types. If comparison and cost are zero, the page is inside out regardless of how good the individual answers are.
One question, one answer. Do not bundle two questions into one entry, and do not split one answer across two.
Write answers to stand alone. Assume the reader sees the answer and nothing else on the page. No "as described above", no "this range" without saying which range.
Put the answer first. The first sentence should answer the question directly. Detail follows.
Ten to twelve questions on a category page, five to seven on a product variant page. More than that and the set stops being read; fewer and the category is under-covered.
A note on the plan to reduce FAQ text to 10 or 12 point. We would advise against it. Text sized so that it is technically present but not meant to be read is the pattern search engines treat as hidden content, and the risk is that the section is discounted entirely rather than read more eagerly. The FAQ section is also genuinely useful to the buyers who reach it. The better route to the same goal is fewer, sharper questions at normal size: a set of ten that answer what buyers ask will outperform a set of twenty that restate the brochure, at any point size.
Review the set, not the question. These guidelines work at the level of a page's whole FAQ set. A page can contain ten well-written answers and still fail, if all ten answer the same kind of question.
The page by page tables that apply all of this are in the FAQ set for each Water Coolers page.