on this page
Executive summary
The AI search visibility category in India is roughly eighteen months old, has no certification, no accepted measurement standard, and a growing number of SEO agencies who changed the word on their homepage and nothing else. That means the buyer carries the entire burden of due diligence. These eleven questions are the ones I would ask if I were on the other side of the table, grouped into four areas: can they measure, do they understand the mechanics, are they honest about outcomes, and are the commercials clean. Each question comes with the answer that should reassure you and the answer that should end the meeting. Score the provider out of 22 and decide from the number, not the pitch.
First: do you need a partner at all?
Before the eleven questions, settle a cheaper one. A monitoring tool and a person who will act on it can cover a lot of ground for a small practice.
A monitoring tool tells you that you are invisible. It does not make you visible. That is the whole distinction. Tools report. Partners execute. If nobody in your firm is going to write the FAQ architecture, fix the schema, chase the reviews and post on Reddit, a dashboard is a ₹2,600 monthly reminder of a problem you are not solving.
What the category actually costs
Published tool pricing, verified September 2026:
| Tool | Entry price | Converted (approx.) | What it does |
|---|---|---|---|
| Otterly.AI | $29/month | ₹2,600 | Monitoring, 15 prompts, 4 engines |
| Peec AI | $95/month | ₹8,400 | Monitoring and analytics, 50 prompts |
| AthenaHQ | $295/month | ₹26,000 | Self-serve platform |
| Profound | $399/month | ₹35,000 | Enterprise analytics, white-label |
Sources: vendor published pricing via 2026 tool pricing roundups. Converted at roughly ₹88 to the dollar. Every one of these is a reporting product.
Done-for-you work in India is a different market with far less price transparency. Indicative bands from what I see quoted in Mumbai:
| Provider type | Typical monthly | What you usually get |
|---|---|---|
| Freelancer or small SEO shop | ₹8,000 to ₹20,000 | SEO deliverables with AEO vocabulary |
| Generalist digital agency add-on | ₹20,000 to ₹40,000 | A tool licence plus a monthly report |
| Specialist done-for-you | ₹18,000 to ₹50,000 | Technical, content, entity and citation work, measured |
Who is actually selling this in India?
Four different things are being sold under the same three letters. Knowing which one is in front of you answers half the questions before you ask them.
Part one: can they measure?
If a provider cannot measure, nothing else they say can be checked. Start here.
01. What exactly are you measuring, and can you show me the rubric?
Good answer. They describe a defined scoring system. Something like: named as top recommendation scores three, mentioned scores two, cited as a source scores one, absent scores zero, aggregated across a fixed query bank into a single number with a stated maximum. They show you the sheet.
Bad answer. "We track your AI visibility." "We use an industry-standard score." There is no industry standard. Nobody has defined one. If they claim otherwise they are either repeating a vendor's marketing or inventing it.
02. How many queries, on how many platforms, and how often?
Good answer. A specific number of conversational queries, tested across at least four or five platforms, on a fixed cadence with screenshots archived. They can explain why the number is what it is.
Bad answer. A single query, or three. Or "we monitor continuously" with no query count. Outputs are non-deterministic. SparkToro's January 2026 research found that asking ChatGPT or Google's AI the same question 100 times gives less than a 1 in 100 chance that any two responses list the same brands. A provider who tests one query once and shows you the screenshot is showing you noise, not a measurement.
03. Will you show me my baseline before I sign anything?
Good answer. Yes, either free or as a small paid audit that credits against the engagement. They will tell you where you stand before you commit to a retainer.
Bad answer. "We do that after onboarding." A provider who will not measure you before taking your money is protecting themselves from a comparison later.
Part two: do they understand the mechanics?
These three questions cannot be answered from a sales deck. They require the person to know how retrieval works.
04. Is my website indexed in Bing, and how would you check?
Good answer. They already checked before the meeting, or they open Bing Webmaster Tools on the call. They mention IndexNow. They explain that ChatGPT's retrieval has leaned on Bing and Copilot runs on it directly, so Bing indexation is a prerequisite, not an optimisation.
Bad answer. A blank look, or "Bing doesn't matter in India." Indexation is the hard gate. If your site is not in the index a platform retrieves from, you cannot be retrieved, cited or recommended, no matter how good your website is. This question alone separates people who understand the pipeline from people who have read three blog posts about it.
05. What percentage of your work happens off my website?
Good answer. Most of it, after the first month. Entity propagation across directories and professional listings, LinkedIn, review velocity, Reddit and Quora, citation building, earned placements. They can name the specific surfaces for your profession: the ICAI member directory, JustDial, Practo, the chamber listings.
Bad answer. "We optimise your website for AI." That is roughly a third of the work. SEMrush analysed 325,000 prompts in March 2026 and found LinkedIn is the most-cited domain for professional queries. SE Ranking, November 2025, found domains with substantial Reddit and Quora presence have around 4x higher citation likelihood, and domains with review-platform profiles around 3x. None of that lives on your website.
06. Which of my competitors is being recommended instead of me right now?
Good answer. They name two or three firms, and can show you the query that produced them. This takes ten minutes to prepare and almost nobody does it.
Bad answer. A generic answer about competitors in your category. If they have not run a single query about your firm before pitching you, they have not done the one piece of work that would have taken the least effort.
Part three: are they honest about outcomes?
The category rewards overclaiming. The buyer's only defence is to test for it directly.
07. Do you guarantee I will appear in AI answers?
Good answer. No. Clearly, immediately, with an explanation. AI outputs are probabilistic systems nobody controls. What a serious provider guarantees is the complete delivery of the work and honest reporting of the trajectory.
Bad answer. Yes. Any version of yes. "Guaranteed ChatGPT ranking" is the single clearest disqualifying signal in this category, and it is being advertised. There is no ranking. There is nothing to guarantee.
08. What can this not do for me?
Good answer. They have a ready list. AI platforms send very little referral traffic, under 1% of referrals for most brands per Similarweb's 2026 index. The work decays without maintenance, since EMARKETER found 40 to 60% of cited sources rotate month to month. It will not fix a practice with poor reviews. It protects referrals rather than replacing them.
Bad answer. A pause, then a reframe into a benefit. A provider who cannot name the limits of their own service has either not run it long enough to find them, or is choosing not to tell you.
09. How does this stay compliant with my professional body's rules?
Good answer. For a CA firm, they know ICAI's revised guidelines, effective April 2026, permit digital presence and factual information about the practice, and they can tell you where the line sits. For an advocate, they know Bar Council Rule 36 is considerably tighter and will say so even though it shrinks the engagement. For a clinic, they know the medical advertising restrictions.
Bad answer. "We'll check." Or worse, a suggestion to publish client testimonials and comparative claims that your regulator does not permit. A partner who does not know your profession's advertising rules will eventually write something that gets you a notice.
Part four: are the commercials clean?
10. Who does the work, and what do I own if I leave?
Good answer. A named person, with a clear answer on capacity. On ownership: your Google Business Profile access stays yours, they hold Manager not Owner rights, the schema lives on your site, the entity brief and query map are handed over, and the content is yours. Nothing is held hostage.
Bad answer. Vague team references, or an admission after probing that the dashboard, the listings or the content disappear when the retainer ends. Ask specifically: "if I stop paying next month, what do I lose?"
11. What is your price, as a single number, and what is out of scope?
Good answer. One number per tier, said out loud, without a discovery call prerequisite. And an explicit out-of-scope list: website redesign, paid ads, social media management, video, whatever it is. Scope creep is how these engagements die.
Bad answer. "It depends on your requirements." It always depends somewhat, but a provider who has run this service before knows what their tiers cost. Refusing to name a number before a call is a negotiation tactic, not a pricing model.
Score it out of 22
Two points for a specific, confident answer. One point for a vague but not alarming answer. Zero for a non-answer or a red flag.
| Score | Read |
|---|---|
| 18 to 22 | They know what they are doing. Check references and proceed. |
| 12 to 17 | Competent but unproven, or honest but thin. Start with a paid audit, not a retainer. |
| 6 to 11 | Probably an SEO agency with new vocabulary. Ask questions 04 and 05 again. |
| Under 6 | Walk. |
Any zero on question 07 is an automatic fail regardless of the total. A guarantee is not a weak answer, it is a wrong one.
The red flags, in one list
- Guaranteed AI rankings or guaranteed placements.
- A "proprietary AI visibility score" they will not explain.
- One screenshot of one query as proof of results.
- No willingness to establish a baseline before you pay.
- Pricing that requires three calls to discover.
- No named person doing the work.
- Deliverables that are indistinguishable from a 2019 SEO retainer.
- No knowledge of ICAI, Bar Council or medical advertising rules for your profession.
- Case studies with no numbers, or numbers with no method.
- Anything that stops working the day you stop paying.
Frequently asked questions
How do I know if an AEO or GEO agency in India is legitimate?
Ask them to show their measurement rubric, produce your baseline visibility score before you sign, and state what their service cannot do. A legitimate provider answers all three specifically and refuses to guarantee AI rankings, because AI outputs are probabilistic and no agency controls them.
Should a small firm buy an AI visibility tool or hire an agency?
Buy a tool if someone in the firm will act on what it reports, since entry monitoring starts around ₹2,600 a month. Hire a done-for-you partner if nobody internally will execute the technical, content, entity and citation work, because a dashboard reports the problem without fixing it.
Is AEO or GEO work compliant with ICAI rules for Indian CA firms?
Yes. ICAI's revised guidelines, effective April 2026, permit member firms to maintain a digital presence and publish factual information about their practice, which is what AEO and GEO work consists of. Advocates face tighter restrictions under Bar Council Rule 36, so the scope of permissible work is narrower.
Where to start
You do not need a partner to answer question three. Open ChatGPT and Perplexity, ask what a prospective client would ask, and write down who appears. That is your baseline, roughly. It costs nothing and it changes the conversation with every provider you meet after it.
If you want the proper version, Lumicite runs a fixed-scope AI Visibility Audit for ₹9,500: 25 test queries across five platforms, scored on the rubric described in question one, competitor intelligence on your top three, a technical foundation audit including Bing indexation, and the list of domains AI is currently citing in your category. Five to seven working days, creditable against a full engagement.
Then ask us the same eleven questions. If we score badly, do not hire us.
Sources
- SparkToro, consistency of brand recommendations across repeated AI queries (2026)
- SEMrush, analysis of 325,000 prompts identifying the most-cited domains for professional queries (2026)
- SE Ranking, citation likelihood for domains with community presence and review-platform profiles (2025)
- EMARKETER, month-to-month rotation of cited sources across Google AI Mode and ChatGPT (2026)
- Similarweb GenAI Brand Visibility Index, AI platform referral traffic as a share of total referrals (2026)
- ICAI, revised guidelines on digital presence for member firms, effective April 2026 (2026)
- Bar Council of India, Rule 36 on advocate advertising (2026)
- Published vendor pricing, Otterly.AI, Peec AI, AthenaHQ and Profound, as listed September 2026 (2026)