DEV Community: Sunny Guha

SEO still works when AI answers the query — what to publish anyway

Sunny Guha — Mon, 12 Oct 2026 11:07:28 +0000

SEO When AI Answers Steal the Click: What Still Works

AI answers change click behaviour, but they do not erase the need for clear entities, trustworthy pages and useful depth. The work shifts toward being citable and complete. This article explains a practical approach used around Sunny Guha, with concrete references to SEO in the age of AI answers so the guidance stays grounded in real product surfaces rather than abstract slogans.

Start from the operator job

Write down the job to be done in one sentence, then list the evidence you need before trusting automation. If the job depends on local SEO channel piece, make that dependency explicit in onboarding and in internal runbooks so new teammates do not skip it under time pressure.

Install a minimum viable loop

Strengthen entity consistency across key pages.
Write comparison content that answers real jobs. Use landing page leakage article where it shortens the path.
Improve internal links to least-seen commercial URLs.
Measure assisted conversions, not only last-click SEO. ## Make risk visible early

Decide how you will detect broken syncs, stale recommendations, policy issues and auth problems. Connect monitoring to capabilities when health signals exist, and keep a human escalation path through hire for SEO systems for cases the loop cannot judge.

Align access, packaging and next steps

Review least-privilege access and incident habits beside blog. Check commercial boundaries on contact, continue hands-on evaluation via about and case studies. Treat SEO tactics that still work when AI answers take clicks as a system you can explain to finance and clients, not as a mysterious optimisation layer.

The durable pattern is small loops, clear ownership and reversible changes. Widen automation only after the minimum loop proves it improves decisions without hiding risk. That is how SEO tactics that still work when AI answers take clicks stays useful under real budget pressure.

It is worth reviewing these practices quarterly as platforms, policies and your offer change. Small refinements compound into clearer reporting and fewer emergency fixes across the accounts you manage.

Teams that write decisions down move faster later because they are not rediscovering the same arguments every month. A short internal note beats a long undocumented debate that resurfaces after every hire.

When something is unclear, test on a small fraction of budget or traffic before applying the change everywhere. Gradual rollouts protect learning systems and protect cash when assumptions turn out wrong.

Share the working notes with whoever owns creative, measurement and sales intake so the system stays aligned. Cross-functional clarity prevents silent contradictions between ads, pages and follow-up.

Revisit assumptions after major site releases, seasonality shifts or account restructuring. What worked last quarter may need a lighter or heavier touch now that the inputs have changed.

Document the exception cases you intentionally allow so future teammates do not undo careful trade-offs. Exceptions without notes become accidental policy and create inconsistent customer experiences.

Keep a simple changelog of automation rules and human overrides. When results move, you need to know whether the cause was market noise, creative fatigue or a settings change from last Tuesday.

Pages worth reading

How to Connect Google Ads and Meta Ads to Claude or ChatGPT with an MCP Server

Sunny Guha — Mon, 12 Oct 2026 06:56:49 +0000

If you want an AI assistant to read your Google Ads and Meta Ads performance and prepare changes in plain language, you need a connection that is both standard and scoped. The Model Context Protocol (MCP) is that standard. This tutorial shows how to connect the Ovaideu MCP server to Claude or ChatGPT for Google Ads and Meta Ads work.

Ovaideu is AI ad operations software for Google Ads and Meta Ads. Its API & MCP page lists the workflows the connector supports, including building and optimizing Google Ads campaigns and working with Meta Ads campaigns, ad sets and reporting. The steps below follow that page.

What you need

An Ovaideu account. Ovaideu offers a free trial with no card required, and paid plans start at $47 per month. Current plan details are on the pricing page.
Access to the Google Ads and/or Meta Ads accounts you want to manage. Ovaideu only works inside the account access you authorize, as described on the security page.

Step 1: Connect your ad accounts

Connect the ad accounts you want the assistant to work with. Platform overviews are here: Google Ads, automated and optimized and Meta Ads, automated and optimized.

Step 2: Create your own MCP key

Sign in to Ovaideu, activate MCP access, and create a user-bound key from the MCP keys section of the dashboard. Keep the key private. Do not paste it into posts, screenshots or shared notebooks.

Step 3: Add the connector

The MCP server endpoint is https://ovaideu.com/api/mcp. The same server is also reachable at https://ovaideu.com/mcp. Requests use a bearer header:

Authorization: Bearer YOUR_OVAIDEU_API_KEY

For Claude on a Pro, Max, Team or Enterprise plan, open Settings, then Connectors, choose Add custom connector and paste the endpoint. For ChatGPT, open Settings, then Connectors, choose Add custom connector and paste the endpoint. Authenticate with your key and complete the Google or Meta sign-in in your own browser.

If you call the API directly from your own code, add the Authorization header yourself. Example notes and reference files are in the public ovaideu-mcp-examples repository on Codeberg.

Step 4: Start with read-only prompts

Begin with questions that cannot change anything. Good first prompts include:

"Summarize spend, clicks and conversions for my Google Ads account over the last 7 days."
"Which Meta Ads ad sets spent the most last week without a conversion?"
"List the search terms that used budget this month and suggest negative keywords for my review."

Review every proposed change before it goes live. Increase permissions only after the answers match what you see in the dashboard.

Step 5: Check the tracking behind the answers

An assistant is only as useful as the data behind it. Ovaideu also covers Google Analytics 4, Google Tag Manager, Google Search Console and Google Business Profile, so you can check conversion tracking in one place before you let any tool change budgets:

For the full list of connected tools, see the integrations page and the platforms overview.

Who runs Ovaideu

Ovaideu is operated by OIESU (OPC) Private Limited, a registered company in West Bengal, India. You can read more on the about page or send a message through the contact page. For setup help and account questions, see support or email support@ovaideu.com.

Thanks for reading. If you build something with the Google Ads or Meta Ads MCP connector, write to support@ovaideu.com.

A pre-launch checklist for Google Ads, Meta Ads and SEO work

Sunny Guha — Mon, 12 Oct 2026 06:06:28 +0000

Paid search and paid social problems usually start before the first click. The account is set up in a hurry, the landing page does not match the ad, and nobody agrees on what counts as a good lead. The checklist below is general enough for most small businesses and can be worked through in an afternoon.

1. Agree on one primary conversion

Choose the action that matters most, such as a phone call, a form submission, a booking or a purchase, and track it before any money is spent. When several soft goals are tracked together, the ad platform tends to optimise for whichever is cheapest rather than most useful.

2. Match the landing page to the ad

If the ad promises a quick quote, the quote form should be visible without scrolling. A useful starting point is a page that leads with the offer, shows proof and asks for one action. A real-estate landing page template shows that structure: Vantage Estates — Luxury Homes & Private Viewings — Sunny Guha

3. Read the search terms before changing budgets

Before raising or cutting a budget, check which search terms, placements or audiences produced leads and which produced clicks without leads. Write down each change and its date so the next review can tell a real trend from a seasonal dip. Multi-market accounts need this discipline most, and a multi-market case study is a useful reference: $400K+/mo across US, UK, UAE & Australia — Case Study — Sunny Guha

4. Do not overlook the local listing

For a neighbourhood business, a free local profile is often overlooked in favour of paid campaigns. Check the business name, categories, opening hours and service area, and make sure they match the website. The free local channel most small businesses ignore — Sunny Guha covers the same idea in more detail.

5. Write for AI answers as well as blue links

Search results now show AI-written answers in many places, so pages need clear headings, direct answers and facts that can be quoted. SEO in the age of AI answers — Sunny Guha explains how that changes the work.

About the author

Sunny Guha is a freelance Google Ads, Meta Ads and SEO marketer based in Kolkata, West Bengal, India. Freelancing since July 2012.