Hi there!
I’ve been working on CandleDojo, a browser-based chart-reading trainer for traders.
The product idea was always pretty clear in my head:
help traders get structured reps reading historical charts, recognizing candlestick patterns in context, and improving chart-reading skill without turning practice into random screen time.
But I realized something uncomfortable:
the product made much more sense when I explained it than when the website explained it.
That is a problem for normal SEO, and it is also a problem for GEO.
If someone searches Google for a candlestick practice tool, a chart-reading trainer, or a way to practice price action, the site needs to make that obvious.
And if someone asks ChatGPT, Claude, Perplexity, or Google AI something like:
what’s a good tool for practicing chart reading?
then CandleDojo needs to be understandable enough to be suggested as an answer.
So I started rebuilding the public side of CandleDojo around those goals.
Why I cared about SEO and GEO
At first, most of the discoverability was basically brand-driven.
That means if someone already knew CandleDojo or typed Candle Dojo, they could probably find it.
But that is not the real challenge.
The real challenge is showing up for the kind of searches people use when they do not know your product yet:
candlestick practice toolchart reading trainerhow to practice price actiontrading replay simulatorbar replay vs paper tradingbest way to practice chart reading
That is the SEO side.
The GEO side is slightly different.
For me, GEO is not about stuffing “AI” everywhere.
It is about making the site easier for generative systems to classify, summarize, and recommend.
In other words:
- what is this product?
- who is it for?
- what problem does it solve?
- when should it be recommended instead of something else?
I wanted CandleDojo to answer those questions much more clearly.
What CandleDojo actually is
This was the first thing I had to fix.
I stopped assuming the branding alone was enough.
So instead of leaving the product too abstract, I started being much more explicit about the language on public pages.
CandleDojo is now described more clearly as:
- a chart-reading trainer
- a candlestick practice tool
- a product for price action practice
- a way to train with historical charts and structured reps
That sounds simple, but I think it matters a lot.
Search engines do better when the product category is clear.
LLMs also do better when the product can be placed into a clean mental bucket.
If the site only sounds clever or branded, it is harder to rank and harder to recommend.
If the site clearly says what it is, the odds get much better.
What I changed on the site
The biggest shift was structural.
I stopped treating public pages like isolated marketing pages and started treating them like a real content system.
I split the public surface into clearer sections:
-
/guidesfor educational and problem-aware content -
/patternsfor glossary-style candlestick pattern pages -
/vsfor comparison intent -
/toolsfor interactive assessments -
/insightsfor first-party research and reporting -
/newsletterfor owned audience growth
That gave each page type a more specific job.
For example:
- a guide can target questions like “how to practice price action”
- a comparison page can target high-intent searches like “CandleDojo vs TradingView Replay”
- a tool page can capture people looking for a chart-reading assessment or candlestick quiz
- an insight page can build trust and give the product something more citeable
That last point matters more than I expected.
A lot of GEO-friendly content is not just “content.”
It is content that is easy to quote, summarize, or reference.
What I changed specifically for GEO
For GEO, I tried to focus on clarity over gimmicks.
A few things seemed especially important:
1. Clear product classification
I made sure the pages say what CandleDojo is in plain English.
Not just “train your chart reading.”
More like:
- CandleDojo is a chart-reading trainer
- CandleDojo is a candlestick practice tool
- CandleDojo helps traders practice reading price action on historical charts
That gives LLMs much better raw material.
2. Pages that match recommendation intent
A lot of AI questions are recommendation-shaped.
So I leaned into content that naturally answers those kinds of prompts:
- best way to practice chart reading
- what makes a good chart-reading trainer
- bar replay vs paper trading
- candlestick pattern practice tool
- trading replay simulator
Those are not giant vanity keywords, but they are much closer to actual product discovery.
3. Better internal linking
I wanted the public content to feel like a connected graph instead of scattered pages.
So pattern pages can point to related tools or guides.
Guides can point to assessments.
Insight pages can point back into the product.
Comparison pages can link to deeper educational content.
That helps both users and machines understand how the topics connect.
4. Crawl accessibility for public pages
I also updated the crawl policy so public content can be accessed more broadly, while private app routes stay blocked.
That means pages like guides, tools, comparisons, and insights are part of the public surface, while routes like /play, /profile, /auth, and /api stay out of that layer.
For me, that is the right split:
make the public knowledge surface accessible, and keep the actual app internals private.
What I changed for measurement
One thing I really did not want was to spend weeks “doing SEO” and then only know whether pageviews went up.
That is not enough.
So I added more attribution and event tracking around the public funnel.
I started tracking things like:
- page views
- CTA clicks
- tool starts
- tool completions
- newsletter signups
- signup starts
- signup completions
I also added first-touch attribution on signup.
That matters because I want to know which content types actually help create users.
Not just:
- which page got traffic
but:
- which page led to a newsletter signup
- which page led to a tool completion
- which page helped start an account signup
That feels much closer to the real job.
What I changed for scale
I also built a weekly content pack workflow around CandleDojo.
Instead of thinking about SEO posts, videos, newsletter content, and directory snippets as separate workflows, I wanted one source topic to create a full pack.
So one topic can produce:
- one primary content asset
- one YouTube script
- a few short-form hooks
- one newsletter block
- one directory/profile snippet
This makes the system much more scalable.
It also fits how I think about search now.
The same core idea should be able to show up across:
- Google Search
- AI-generated recommendations
- YouTube
- newsletter
- directory/referral traffic
Different channels, same underlying intent.
That feels much more durable than creating random one-off content.
What I learned
A few things became clearer while doing this.
Broad keywords are not the point
Trying to rank for something like candle, dojo, or even trading is not the right game.
The better targets are narrower and more product-shaped:
- chart-reading trainer
- candlestick practice tool
- best way to practice chart reading
- trading replay simulator
- bar replay vs paper trading
Those are much more realistic and much more useful.
GEO is not magic
Just letting AI crawlers access your site does not automatically make your product recommendable.
If the pages are vague, thin, or hard to classify, LLMs still will not have much to work with.
The real work is still:
- clarity
- structure
- useful page types
- internal linking
- recommendation-friendly language
Tools and comparisons are strong assets
I think interactive tools and comparison pages are underrated.
A generic blog post can be useful, but a good tool page or a strong comparison page often does a much better job of matching high-intent discovery.
The content system matters as much as the writing
The more I worked on this, the more it felt like an information architecture problem, not just a writing problem.
You are not only writing articles.
You are teaching search engines and LLMs how your product should be understood.
Where I am now
I do not think SEO or GEO is something you “finish.”
It feels more like building a long-lived public knowledge layer around the product.
For CandleDojo, that layer now has a much clearer shape:
- educational guides
- pattern pages
- comparison pages
- interactive tools
- insight pages
- newsletter capture
- attribution so I can measure what actually works
It is still early, but this already feels much better than hoping the homepage alone will do all the work.
Final thought
My current takeaway is pretty simple:
if you want your product to show up in both search and AI recommendations, do not just publish more content.
Make your site easier to understand.
Make it obvious what the product is, who it is for, and when it should be recommended.
That has been the real shift for me with CandleDojo.
If you want to check it out, it’s here:
Top comments (0)