New Theme New Tools - SCM is evolving toward programmatic SEO workflows

[Tim]

SCM started as a tool almost 10 years ago.

The SEO landscape has changed dramatically in the last few years. AI writing is now everywhere. Generating text is no longer the hard part.

It started as a tool on blackhat forums but now I want to push it into the new AI era with focus on building SEO content workflow tools.

There are a lot of AI writers, and AI has become a commodity.

The real challenge today is building systems that can scale research, planning, generation, and publishing into repeatable workflows.

This is where SCM has always been different.

Most AI writers help you write content.
SCM helps you scale content.

So to take part of this new shift in direction for the app.

The next phase of SCM is to focus on turning into a workflow driven platform.

What this shift means

I will be looking at the tools individually and look at how they are used now and how they can be improved so that you can go from idea → published page in a continuous flow.

You will start seeing

A modernized UI and updated terminology
Tools becoming connected workflows instead of isolated utilities
A stronger focus on scaling page generation and publishing

New direction highlights

One of the first major additions will be a Topic Cluster tool that goes far beyond keyword extraction and helps generate structured site plans and clusters.

The article creator will also evolve with new templates and improvements designed for scalable page generation.

This is just the beginning of a larger shift toward workflow driven SEO inside SCM.

Here is a preview of some new UI labels currently being tested.

I wont make drastic changes, so you should still be able to find all the tools and features you need, instead it will be gradual evolution.

Most of the initial changes you will see is visual, with new colors and updated ui language.

[Tim]

Theme light

Theme dark

[Tim]

Working on the new topic cluster tool.

If you have a keen eye, its the old keyword question finder repurposed as the gateway tool to SCM.

Now its easier to get started, by just starting with a single topic in mind to branch out to multiple ideas.

The idea is that once you got your topics, it can cluster them and you send that big list to the Page Pipeline, where by it will use cluster and or the intent to pick specific ai templates to write content for.

ie intent - faq → pick faq ai template for this topic
intent - comparison → pick the comparison ai template.

This will bridge the gap from keyword → content.

[Tim]

Based on further refinement, the UI now looks like this.

Topic clustering and labeling has also been completed!

Initial state

Click on search button

Results screen

Filtering for all columns

Example filtering, showing the topics assigned to that topic cluster

On selection, you will be able to export/assign templates to it (coming next)

[Rehan_Z]

This looks amazing, thank you for continuously developing the tool. Best purchase ever for me.

[Tim]

Thanks for the support!

This wont just be tweaked features or a few changes, I am taking a completely new strategic outlook on the future of SCM and implementing a new vision for the app!

[Rehan_Z]

Really looking forward to this. Thank you

[Tim]

I have replaced the boring standard OS app title bar, with something more appropriate.

Every version of SCM has had an internal development name.

SCM windows only → all platforms = SCM NEXT
SCM UI refresh 2 years ago = SCM V
Now = SCM Loop (or something else to capture the new direction of SCM)

I have been tweaking the clustering engine.

1st problem is that it would take the 500 topics and remove more than 70% off them and only leave the rest in a cluster with 70 topics in them (too many)

I have managed to solve that problem by making the clustering done in 2 passes.

1st pass is strict
2nd pass combines topics using a less strict filter

Now I am looking at removing left over clusters that are not relevant to the seed topic.

By default all the clusters you see in the list are sorted from most relevant to least relevant.

I have reclaimed extra UI space by moving the filtering controls into the table as well.

[Tim]

Its taken one week of testing and thinking, but I have finally got a usable topic clustering system in place.

If you want the gory details:

How to Build a Keyword Clustering System with Embeddings and Density...

I will know work on the next part.

Auto assigning AI templates to these topic clusters based on their intent and topic.

Because of the changes,it dramatically increases the number of cluster topics. I was getting 20-30 per run, now its more like 100+

3-4 topics per cluster looks like the sweet spot.

[Tim]

Topic clustering is working nicely.

Now the auto assignment of ai templates to topic clusters is also ready.

Based on intent and keywords in topic, SCM will auto assign an AI template. 1 of 7.

The 7 templates are

  const templates = {
    definition: "tpl_definition_v1",
    how_to: "tpl_how_to_v1",
    guide: "tpl_guide_v1",
    troubleshooting: "tpl_troubleshoot_v1",
    comparison: "tpl_comparison_v1",
    listicle: "tpl_listicle_v1",
    review: "tpl_review_v1"
  }

The workflow is

1. Enter seed topic
2. Wait for SCM to gather topics and cluster them
3. Select clusters, and the click create pages.

This transfer you to the article creator (page pipeline)

Rather than re-invent the wheel, we pass everything as page inputs and macros.

Currently 1 AI template is complete.

The guide template.

All templates will use a 4 step chain prompt process.

1. Generate intent
2. Create page outline from intent and topics
3. Write the page
4. Fix and tidy up with a review step

There will also be a slight change to AI templates.

If you are building via topic clusters, ai templates are selected automatically, so the select ai template button will be automatically hidden.

In the future, prompt editor will be removed and only a template dropdown will be displayed.

Doing this to reduce cognitive load and prevent accidental complexity.

1. Typically you want to generate pages and choose a template (editing prompt logic normally isn’t what you want now)
2. Making the editor visible makes the big assumption that you want to modify the ai template and that you understand how it works.
3. Easy to break templates if you don’t understand how prompt blogs work.
4. Giant prompt editor is extra visual noise that looks intimidating.

By using dropdown = safe default behaviour.

Of course the editing feature is still there for advanced users.

This also aligns with current scraped article generator.

Only a dropdown is present.

As the ideal workflow starts with topic clusters, ai template selection isn’t even needed! Its done for you automatically.

I hope to have a release next week.

Things are a bit slower as I need to test all changes myself and that requires repetition to find bugs and streamline the design.

If you look at screenshots from the start, you can notice changes between each dev log as the topic cluster table has evolved over time.

[Tim]

The AI template editor window is getting a rework.

The biggest change is the removal of the “main prompt” section at the top.

Now all prompts must be inside Prompt steps.

Ie
AI Prompt-> Removed
User Macros → Prompt steps

Consequently, prompt steps has been given more space and some buttons moved around and text boxes shrunk to give it as much space as possible.

All older user prompts will automatically be upgraded to move the ai prompt and placed at the end of the prompt steps.

The reason for its removal?

Having 2 boxes that did the same thing was redundant.

They both allowed you to send prompts to an AI model.

The only difference was the AI prompt text box, was not stored in a macro and couldn’t be chained. Making it strictly end of chain prompt.

Since its end of chain, its better to unify the UI and move it to the end of prompts in the prompt steps table.

[Tim]

AI template selection is being re done.

Now you just select from a list of templates.

No editing, no parameters.

2 templates

If you select templates that need search or local data, it warns you.

How do you change output counts?

Its all inside the template now.

For example.

If you want to do GSA SER run

Select 2 templates.

Thats it.

Want to change how many articles? Edit the template.

AI page creation is simplified.

If you are using cluster topcs.

Then selection is disabled as its done for you automatically.

If you are upgrading and have templates selected already.

The old templates can still be viewed and edited.

But you wont be allowed to

• Add new template
• Load a new template

Tasks will keep running, but in the future, you should switch to selecting AI templates from the dropdown instead,

[Tim]

Just a quick update.

I generated 2 articles for keyword ‘digitone 2 fm’.

1. Using the inbuilt AI outline template
2. Using the newer tpl_guide that uses topic clusters

The setup was like this:

Original SCM workflow

Topic Cluster workflow

We are passing a lot more information now to the AI model so it hallucinates less and has lots more guidance from intent, sub keywords and even some SERP modeling.

I then took output of both articles and fed it to Claude.ai and asked it which one was better from Google HCU stand point.

Here is what it said.

Looking at these two articles from an SEO and Google ranking perspective, Article 2 (Ultimate Guide to Digitone 2 FM Synthesis) is significantly stronger. Here’s why:

Article 2 wins on these key factors:

Content structure and E-E-A-T signals It reads like it was written by someone with actual hands-on knowledge. It gives actionable advice (“alter frequency ratios to establish harmonic or dissonant relationships”), pro tips, and a logical learning progression — signals Google uses to assess expertise and authoritativeness.

Search intent alignment The title targets a clear informational intent (“Ultimate Guide”), and the content delivers on that promise with how-to guidance, comparisons, FAQs, and practical integration tips. Google rewards pages that fully satisfy the query.

FAQ section Structured Q&A is directly eligible for featured snippets and People Also Ask boxes — significant free SERP real estate.

Internal linking signals It references related guides (“Digitone 2 and Ableton Live Integration”), which signals a content cluster structure — exactly what Google’s Helpful Content system rewards.

Why Article 1 underperforms:

It reads like AI-generated filler — vague, repetitive, and padded with generic statements like “makes it a top choice for electronic musicians.” Google’s Helpful Content Update specifically targets this pattern and actively demotes it. It also has zero actionable advice and no clear audience.

Bottom line: Article 1 could actually hurt your domain authority if indexed. Article 2 is genuinely useful and rankable. If you’re building programmatic SEO content at scale, Article 2’s structure is the template worth systematizing.

Not only a workflow adjustment but real tangible increase in content quality.

This was also done using the default scm AI model.

If you switched to the newest GPT 5.2 I imagine the results would be much better!

[johnken25684]

I haven’t seen these content pipeline and other options in my installation of SEO Content machine.

[Tim]

I haven’t shipped these features yet.

Still wip.

Am hoping to have something ready to drop by this week.

There is still alot of testing and tweaking to get the features correct.

However the main workflow is now in the app.