We read 29 SEO products before building ours. Here is the gap we chose.

On 3 November I set a rule for myself: research existing software and best practice before designing any module. My diagnosis of the week had been plain. We lacked upfront planning and research, and the key research that helped came from ChatGPT and Gemini reports, not from the build itself.

SEO was the first module to get the rule. This is what the research found and what we decided.

Why we did it

IIPTA, Mukika and the YuktiBOS site all need SEO. Results take months, so I said start now, not next week. But I did not want Claude running audits for us by hand. The IIPTA team needs to run SEO themselves, inside YuktiBOS, with the same pattern as content: data, a page audit, recommended changes with exact edits, a human review, then apply the change (through GitHub when connected, otherwise as a task for a person).

Before designing that, we wanted to know what already exists, what it costs, and where it falls short.

How we did it

A research agent ran about 16 web searches and about 25 page fetches, following the way a competitor study is normally done: list the field, profile each product on the same fields, build a matrix, find gaps, then propose.

The limits, stated in the research itself:

• Most pricing comes from review and roundup sites, not vendor pages, which load with JavaScript and could not be read by the fetch tool.
• Several roundups are written by competitors. They were flagged.
• Prices conflict between sources. For example, one tool is listed at $49, $99 and $182 per month in one article and $89 to $299 in another. Where they disagreed, both are given.
• Review sites mostly did not load. Weaknesses come from roundup authors and a few snippets.
• Many "best AI SEO tools" pages are affiliate listicles. We used them for prices and features, not opinions.
• Jasper, Moz and Writesonic were researched thinly.
• Rupee figures use about Rs 85 to the dollar.

So every price is "list price around Oct and Nov 2026, verify before quoting".

What we found

The field: 29 products in six groups.

1. Content optimisation: score a draft against search results (Surfer, Clearscope, MarketMuse, Frase, NeuronWriter, Scalenut, Writesonic, Jasper).
2. AI agents that do the work: Alli AI, OTTO, SEO.ai, Byword, Koala, Outrank, SEObot, InLinks.
3. Audit and rank suites: Semrush, Ahrefs, SE Ranking, Moz Pro, Ubersuggest, Screaming Frog.
4. AI-search visibility: Profound, Otterly, Peec, AthenaHQ, Scrunch, Semrush AI Visibility, Ahrefs Brand Radar.
5. Plugins that write to the site: Rank Math, Yoast.
6. Open source and data: OpenSEO, DataForSEO.

Table stakes. A crawler-based technical audit; on-page checks; query data and position tracking; Core Web Vitals; a content brief or editor score; a shareable report; AI-search visibility reporting (even the big suites ship one now); schema generation.

Differentiators. Only a few tools change the live site: Alli, OTTO and InLinks inject a script, and the WordPress plugins write to the database. The script route has a trap. Stop paying and the fixes vanish. Credits are the most complained-about pricing model.

A caution on AI-search tools. A quote from an industry article captures the problem: if you use three tools and give them the same prompts, you get three different answers. Agency executives call them "a benchmarker", not a source of truth.

Price bands (rounded): free tools such as Search Console; budget tools at $9 to $50; SMB at $59 to $130; agency tools at $150 to $500; enterprise from $500 to $1,500 and above. Indian roundups list the same Western tools at roughly Rs 3,000 to 10,000 a month, with agency tools at Rs 20,000 and more.

The gap we chose

Eight gaps came out of the matrix. The ones that shaped the product:

1. The full loop in one place with a human approval step. Suites diagnose but do not fix. Agents fix but hide the audit trail and lock fixes behind a subscription. Nobody makes "this fix was verified on re-crawl, here is what happened after" the main screen.
2. Plain language for non-experts. Almost everything assumes you know what a canonical tag is.
3. Done-for-you and do-it-yourself on one platform, so an agency like IIPTA or Mabbit can use it as a workbench and a client can use it themselves.
4. Indian small businesses and education businesses. I found no tool built for institutes (course pages, admission funnels, Course and FAQ schema, near-me search, multi-campus pages). I could not prove none exists. It is a gap in what I could find, and the research says so.
5. Honest AI-search measurement. Ask each question several times, show the range, and tie the result to what to fix on the page.
6. Fixes that persist without a subscription, as pull requests or CMS writes the owner keeps.
7. Predictable cost: show an estimate before each run.

The positioning line we wrote: the SEO tool that finds the problem, writes the fix, waits for your OK, applies it, and checks it worked, in plain language, priced for Indian businesses.

The design rule that keeps it affordable: code finds facts, AI only explains and words, a human approves.

Then we critiqued our own spec

A separate reviewer agent read the SEO spec (122) and found must-fix problems before any code. Three examples:

• It proposed building Search Console and rank tracking. They already existed in the code.
• It proposed a new "vitals" table that contradicted its own line saying to extend an existing one.
• AI wording cost was unbounded: a 500-page audit could cost about Rs 2,500 per site at the target of Rs 5 a page. The fix was to word only pages with an issue and high inlinks, with an estimate shown first.

The review also said that three specs written the same day (SEO, client section, articles) would take about nine Sonnet days in reality, well above the days the specs stated, and that cuts still left it over one week.

What it changed in YuktiBOS

SEO v1 shipped in the 4 to 5 November build: site audit, change list, re-check, AI-search and Search Console panels, and a checkup engine. The public "SEO Checkup" page is next on the list.

What is still open

• I have not yet verified the audit live on iipta.com.
• The paid tier price range (Rs 999 to 2,999 per month) in the research is an assumption to test, not a finding.
• The public tool needs abuse controls and a DPDP Act review before launch. The research said it had not verified the legal specifics.
• My own list of AI SEO companies has not been added, so the field is incomplete.
• A stand-alone CRM competitor study does not exist in our files. The method note in this research refers to one as the model to follow.

Numbers at a glance