Marketing teams are shipping more content than ever with AI. New blogs, landing pages, and guides go live every week. Then they open Search Console and see the truth: most of those URLs barely get any impressions.

The issue is not “Google vs AI.”
The issue is AI content that is written without real search data.

Search has changed faster than content workflows

Several shifts are hitting at the same time:

• Nearly 60 percent of Google searches now end without any click to a website, thanks to answer boxes, AI summaries, and richer SERPs.

• Studies show AI based overviews and summaries now appear on a growing share of queries, especially longer, question style and informational searches, which are the same ones many blogs target.

• At the same time, total search volume has grown, and organic search still drives more than half of all website traffic and converts far better than outbound, which means SEO is not dead, but the bar has gone up.

So users are still searching, but they are:

• Asking more complex, conversational questions.

• Getting many answers directly inside the SERP.

• Clicking through only when they see something that matches their intent clearly.

Traditional content workflows have not kept pace.

The old SEO model that is breaking

For years, the routine looked like this:

• Pick a keyword with good volume.

• Write a long blog post about it.

• Put the keyword in H1s, H2s, and meta tags.

• Build a few links and wait.

That keyword → blog post → ranking model is not holding up now.

Search behavior around a single topic splits into many different intents. A broad term like “marketing automation” hides dozens of real questions:

• “marketing automation examples”

• “marketing automation workflows for ecommerce”

• “marketing automation vs crm”

• “how to calculate marketing automation roi”

Each one needs its own angle and format. A single 2,000 word article rarely fits all of them well enough to win a modern SERP that also has AI summaries, video, and product blocks.

On top of that, results pages are crowded with:

• Product cards and shopping units.

• Video carousels.

• Answer boxes and AI overviews.

A generic “blog post” is often the wrong format for what that page is trying to serve.

Why most AI content fails

Most AI content fails for a simple reason: it guesses.

When teams lean on AI without data, the model guesses:

• Topics: based on what “sounds right,” not what people actually search.

• Structure: the same familiar pattern on every topic.

• Keyword coverage: loose synonyms, not real query clusters.

• Intent: a generic reader, not a searcher with a job to be done.

Google does not need special AI detection to filter that out.
User behavior is enough:

• The page does not reflect the range of actual queries around the topic.

• The language feels generic, not like what people type.

• People bounce quickly, or rarely click in the first place.

So you get a huge volume and almost no search traffic.

At Thirdi, we try to avoid this mistake by starting from data, not from “AI capability.” Our agents do the same thing in paid media: they read thousands of rows of keyword, creative, and social data and turn that into a short list of actions, not just dashboards. The same mindset can, and should, apply to content.

Search data first, AI second

The fix is to flip the order.

Instead of “ask AI for content ideas,” the flow looks like this.

1. Start with real demand

Use external tools and your own data:

• Tools like Ubersuggest and AnswerThePublic help you see real topics, questions, and comparisons people search around your themes.

• Your own data (Google Search Console, site search, and paid search terms) shows what your users actually type, not what you imagine they type.

• On the paid side, Thirdi’s Keyword Intelligence scans thousands of search terms to find leaks, missed intent, and winning themes across Google Ads, with plain language suggestions to fix them. You can treat that same search term data as raw material for content planning.

2. Group by intent, not keyword

Once you have the queries, tag each by:

• Intent: information, comparison, pricing, template, troubleshooting, etc.

• Stage: early research, evaluation, decision.

Decide which groups belong on one page and which need:

• Their own deep dive guides.

• Calculators or templates.

• FAQ blocks or comparison tables.

3. Then bring in AI

Now AI has a clear brief.

You can ask it to:

• Turn that query map into structures, outlines, and drafts.

• Suggest how to cover each intent cluster in the fewest, strongest assets.

• Propose internal links between related pages.

Set hard rules:

• Every section must either answer a query, explain a decision, or give a concrete example.

• No filler paragraphs that repeat the intro in different words.

AI becomes the worker that speeds you up, not the strategist that decides what to say.

A three step workflow that saves hours

Here is a simple workflow that moves from data to draft without bloat.

Step 1: Map your query space (10–20 minutes)

• Use Ubersuggest for keyword ideas and volumes around a seed topic.

• Use AnswerThePublic to pull questions, prepositions, and “vs” queries.

• Export queries from Search Console and ad accounts.

Drop everything into one sheet.
For each query, tag:

• Intent: info, comparison, product, pricing, template, troubleshooting.

• Who: beginner, practitioner, decision maker.

Thirdi already follows a similar pattern on the performance side: agents scan thousands of data points across Google, Meta, and social to surface clear problem and opportunity clusters instead of leaving you with raw tables.

You are doing the same, just for organic search.

Step 2: Design the content map (15 minutes)

Do not jump to “we need 20 posts.”

Instead, decide:

• One or two “hub” pages to define and frame the topic.

• A set of supporting pages or sections, each focused on a specific intent:

  • “for agencies”

  • “for D2C brands”

  • “pricing and ROI”

  • “common mistakes”

  • “templates and checklists”

For each asset, define:

• The exact queries it must answer.

• The right format: guide, checklist, FAQ, calculator, template, or something else.

Once that is clear, ask AI:

“Given this list of queries and their intent tags, propose the structure and headings for each page, and how they should link to each other.”

This mirrors how Thirdi agents turn noisy performance data into a clear set of “do this next” recommendations for campaigns, not just charts.

Step 3: Draft with guardrails (minutes per piece)

Now you let AI draft, but inside sharp boundaries.

Feed the model:

• The list of queries this page must cover.

• The intent and audience level.

• Style rules: simple language, no padding, clear examples, avoid hype.

Ask it to:

• Write a draft.

• Mark which parts address which queries.

• Suggest FAQ schema or internal links where it fits.

Your work shifts from typing every word to:

• Checking that every section maps to real demand.

• Adding your own data, screenshots, and cases (for example, how a D2C brand used better intent targeting to get 250 percent higher sales with Thirdi’s help ).

• Cutting anything that sounds generic.

This can cut content creation time from hours to minutes, while staying closer to how search actually works today.

The common principle is the same: start from data, not from vibes.AI is the worker. Data and humans set the direction.