Pankaj Gupta | Building Zenyro

I help Indian businesses get found on Google & ChatGPT and convert leads with AI automation. Founder, Zenyro. Sharing real numbers from my own sites.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

Google is rolling out ads inside AI Mode in India.

If you run Google Ads for a local business, three things change:

1. Queries get longer. "Best CA in Mohali for a GST notice reply", not "CA Mohali"
2. Your landing page has to answer that exact question, or you pay for the click and lose it
3. Clean conversion tracking matters more than a long keyword list

Fix the tracking before you raise the budget.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

Before you buy any AI automation, write the process in five lines: what triggers it, what comes in, what decision is made, what action follows, and who owns the result. If you can't, automating it only makes the mess faster.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

A fast test for any service page: read only the first 100 words aloud. If it doesn't say what you do, who it's for and which city or market you serve, neither Google nor an AI answer can place you. Fix that paragraph before touching anything else.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@simontomkins Useful. One thing to add: after fixing the Poor group, wait for the 28-day field data to update before judging it. The report lags, so many people think the fix failed when it simply hasn't refreshed yet.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@jatinprajapat29 Good line. The part that still isn't cheap is support, updates and someone accountable when it breaks. Plenty of people can build a quick tool, far fewer want to be on call for it at 2am.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@OSDDigital The shift in how often AI Overviews show matters, but I'd track it for your own queries. A fixed list of 20 queries, checked weekly with the date, shows whether your pages are being cited or skipped. Industry averages won't tell you that.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@luduvigo Same job, so one owner. I'd give it to whoever owns the content, with a monthly report that has both Search Console clicks and a checked list of AI answers for the same 20 queries. Separate budgets just create two reports saying different things.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@Gadi_Samet For a small business I'd split it: Reddit for being found where people compare options, but only with genuinely useful answers. The ChatGPT side leans more on your own site and third-party reviews, so fix those first.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@tec_safwan Lists like this work best with a filter: pick one repeated task in your week, test two tools on it for 7 days, and keep only the one you still open on day 8. A tool you bookmark but never use costs more than a subscription.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@Onoja_Cee Nice start. Early impressions usually sit on pages 4 to 8, so I'd pick the 3 queries closest to your product and make one page answer each one clearly, with a short FAQ. Moving from page 5 to page 2 is where clicks begin.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@pramodk73 Simplest path: export Search Console queries and pages by API into a sheet or database, then give the agent read-only access to that table. Let it flag pages at positions 5 to 20 with high impressions. Keep the write access out of it.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@aakashgupta Search and AI answers share the same index and the same pages, so splitting them is mostly a product decision. For students it's a real loss: sources shown next to the answer are what teach them to check claims.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@rustybrick Good reminder for anyone running SEO as one person's job. Keep your own traffic and leads data in a place you control, so a team shake-up at Google or anywhere else doesn't change what you can measure.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@kimmonismus The pause argument usually skips what the work actually looks like: clinics, small shops and agencies using AI to answer customers faster. Showing real use cases with numbers will do more to change minds than defending the technology in general.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@theo Quiet on headlines, loud in the workflow. The real change this fortnight is how many teams now run agents on routine tasks. That never trends, but it's where hours get saved and where the next round of news will come from.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

A customer messages your business on WhatsApp at 9:40pm. You reply the next morning. They've already talked to your competitor.

I wrote a 7-step guide to building a WhatsApp AI sales bot, plus what Meta's Oct 2026 pricing changes actually cost in India.

Read it here:
https://t.co/5Kv4FOB9Sz

[Pankaj Gupta | Building Zenyro @iguptapankaj]

https://t.co/T4Pcx7bksl

[Pankaj Gupta | Building Zenyro @iguptapankaj]

https://t.co/T4Pcx7bksl

[Pankaj Gupta | Building Zenyro @iguptapankaj]

Five-minute SEO check: search your brand name plus "pricing" and plus "reviews" in Google, then ask ChatGPT the same. Whatever comes up first is what a buyer reads before visiting your site. If it's wrong or thin, that's the next page to fix.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

A quick test before an AI agent talks to customers: give it the 10 most awkward questions your team gets, like a refund past the deadline, an angry message, or something not in your docs. If it bluffs even once, it needs a human handoff rule before launch, not after.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@KrishnaBuildsAI Usage limits matter more than benchmarks for daily work. I'd run each plan on the same real task for a week and log how often it hits the cap. Whichever lets you finish the job without switching models is the cheaper one.

[Pankaj Gupta | Building Zenyro @iguptapankaj]

@achille610 Evidence-based audits beat a made-up score. The best part is each check linking to Google's own guidance, so a client can see why something is flagged. I'd add an effort column so the fixes get ordered, not just listed.