Objections in AI answers: what buyers hear about you
Buyers ask AI whether you are worth it and the answer repeats objections it finds online. How to see which objections recur and what to do about each.
- Evaluative prompts like “is X worth it” surface the objections buyers already read online.
- Group the objections that keep repeating by category, and note the source behind each one.
- Answer each objection on the page that owns that topic, or correct it where the source lives.
- When the objection is true, the fix is product work and content only follows it.
Buyers ask evaluative prompts and answers repeat what they find
Before buying, people ask prompts like “is X worth it” or “X vs Y problems”. The answer assembles what it finds in reviews, forums and articles, so objections published online come back with your brand attached.
These prompts are worth tracking as their own category. Peec's prompt-selection guide treats brand evaluation prompts such as “Is Tesla worth it?” as a separate group for exactly this reason (their framing): they are asked at the decision moment, and the answer is built from other people's hesitations. The objection in the answer is rarely invented. It is a real argument someone made, quoted without your rebuttal next to it.
See which objections keep recurring
Track the prompts that surface hesitations, then group the arguments the answers repeat. In advisible, the Objections tab under Brand Perception groups recurring arguments into categories (product, pricing, trust, support, comms) and shows the sources behind each.
The prompt shapes that surface them are predictable: “is [brand] worth it”, “[brand] pricing complaints”, “[brand] vs [competitor] problems”, “[brand] reliability”, plus the objections your sales team hears on calls, which are the raw list you already have. Run those prompts daily alongside your category prompts so an objection shows up as a pattern across answers instead of one bad screenshot.
The source column is the useful half. An objection backed by three forum threads is a different job from an objection backed by one stale review site, and only one of those you can edit directly.
What to do for each objection category
Answer the objection where a reader and a crawler can both find it: the page that owns that topic, with the current fact stated plainly. Where the source sits on someone else's site, correct it there if the rules allow.
| Category | What answers tend to repeat | Where to answer it |
|---|---|---|
| Product | Missing features, hard setup, “does not do X” | Docs and a direct “what it does not do” section on the product page |
| Pricing | Expensive, per-seat surprises, tiers that hide the real cost | Pricing page FAQ with a worked total for a real team size |
| Trust | Small vendor, data handling, will they still exist next year | Security and privacy pages with dates, plus press coverage you can point to |
| Support | Slow replies, no phone line, unanswered tickets | Support page stating the hours and response targets you actually meet |
| Comms | Silent during incidents, no changelog, slow public response | Status and changelog pages, and a reply in the thread where the complaint lives |
Write each answer as a claim that can be quoted on its own: “support replies within one business day” travels into an answer; “we pride ourselves on support” does not. The mechanics of writing for extraction are in getting cited.
When the objection is true, the fix is product work
If three answers say support takes days and support takes days, no page will fix it. Fix the product, publish what changed, and let the answers catch up.
Content cannot carry a claim that reality contradicts, and the engines also read the reviews where reality shows up. A rebuttal that the rest of the web disagrees with makes the answer worse, because the next version hedges or drops you instead. Half-true objections are the interesting case: if slow support was real before March and is not anymore, say that with a date. “Response times moved to under four hours in March” is checkable and quotable, and it gives the answer a newer fact to prefer.
The order of operations: confirm the objection with data you already own (tickets, churn notes, sales calls), decide whether it is a product problem or a page problem, then fix in that order. A pricing objection that is really a packaging problem gets solved in packaging, and the pricing page follows.
Common questions
Do objections in AI answers cost sales?
They reach buyers at the moment of decision, which is the worst place to be wrong. Treat a recurring objection as a deal risk and check which prompts repeat it before you prioritise.
Should we reply on Reddit when a thread carries an objection?
Where the thread is active and the rules allow it, reply with the current fact. That reply often becomes the source the next answer cites.
How is this different from sentiment monitoring?
A Sentiment score tells you the tone is worsening. The Objections view names the specific argument and the domains behind it, so someone can go fix the page.
How long until an objection stops appearing?
Weeks to months, and only when the sources carrying it change. A new FAQ on your site rarely outvotes four forum threads by itself.
Brand perception in AI answers: what AI says about you
Visibility tells you whether AI answers mention you. Perception tells you what they say: the attributes, objections and factual claims behind each mention.
Which sources AI engines cite, and what to do about each
Review sites, forums, YouTube, Wikipedia, news, your own docs and listicles each signal something different to an engine. What each does and your move for it.
How to fix wrong information AI says about your brand
AI answers repeat wrong claims about brands. Find the claim, trace the source page feeding it, correct that page, and re-run until the claim stops appearing.
Track this for your own brand
advisible runs your prompts on ChatGPT, Perplexity, Gemini, Grok and Claude every day, and every number shows how many answers it is based on.