What Surfer, Frase, MarketMuse, Scalenut all get wrong

_{Day 175/100}

Hey—It's Tim. 

I have a confession.
I love SurferSEO.

I was one of the first 200 users. I’ve met the founders. Surfer is the reason Penfriend exists.

And I still haven’t logged in for over a year.

Not because I’m mad. Because I stopped trusting what the score was telling me to do.

I used Surfer on everything—my work and 500+ client articles a month. Then AI articles hit. They were… not good. That was the moment I said, “I can do this better,” and started building Penfriend.

I kept the scoring for a while. Then I noticed something: the more I wrote like a human and covered the idea properly, the better I ranked. Not because I sprinkled the “right” phrases across 1,800 words. Because I actually finished the thought.

So I changed the workflow.
Take the pages that rank. Pull them apart with AI. Map the topics, not just the terms. See what’s missing. Add it with intent.

Where the comparison tools go wrong

After dropping Surfer I tried pretty much everything else.

Frase - The Topic Score is a TF-IDF/term checklist; I’ve watched scores jump by adding variants, even when the section isn’t better.

MarketMuse - Their score literally tallies term mentions (points per topic); I’ve had “higher than everyone” scores that still don’t rank.

Semrush (SEO Writing Assistant) - The SEO score feels box-ticky and opaque on why a target changes - surface-level nudges over substance.

Scalenut - The grade pushes an NLP key-term checklist and density - easy to chase numbers instead of finishing a sub-topic.

Clearscope - Content grades can incentivize keyword stuffing; I’ve fought the tool’s grade vs. actual quality more than once.

NeuronWriter - The score rises as you add recommended terms; helpful, but it’s still a term-inclusion proxy—not proof you closed the loop.

They optimize for distribution of words. Google increasingly rewards completion of ideas.

Scorers push you to:

• Chase average term frequency across the whole page.

• Mirror outlines from what already ranks.

• Ship paragraphs that mention everything and explain nothing.

Reality check:

• Agents, summaries, and readers latch onto self-contained sections that fully resolve a sub-topic.

• Topical completeness per section beats keyword coverage per page.

Ranked by paragraph, not page

This part matters. AI Overviews/“AI mode” are effectively scoring paragraphs. Not your whole article.

That changes how you win.

So in Penfriend, when we generate your outline overview, we don’t just list H2s. We detect the ranking paragraphs inside competitor pages, group them under the right sub-topics of your post, and then align each to a focused set of keyphrases.

Result:

• You write to finish a sub-topic inside one section.

• You only care about a few keyphrases per section.

• And because of how Penny writes, the surrounding entities/context are covered by default - so you hit completeness without stuffing.

We’re doing it differently

Penfriend is topic-first.

Start with the concept you want to own. Break it into sub-topics.
For each sub-topic, we show:

• Who covers it on page one.

• How deeply they cover it (down to paragraph-level evidence).

• What’s missing (angles, entities, proof, objections).

Your score isn’t “Did you say the magic words?”
It’s “Did you finish the job on each sub-topic?”

Write section-complete, not page-sprinkled.

The takeaway

These “SEO Tools” aren’t bad.
They’re just measuring the wrong thing.

If you’re optimizing the distribution of words, you’ll always look average.
If you’re optimizing the completion of idea -at the paragraph level- you’ll start looking definitive.

We’re building for the latter.

✌️ Tim "Paragraph Enjoyer" Hanson
CMO @Penfriend.ai

Same brain, different platforms: X, Threads, LinkedIn.

P.S. tiny secret, big lift:
Google filetype:pdf <your topic> and skim the table of contents in the first 3 PDFs. Those TOCs are researcher-grade sub-topic maps your competitors’ SEO tools usually miss.

Steal the best two, turn them into airtight sections (paragraph-complete), and add jump links from your intro.

It feels like cheating because… it kind of is.

Penfriend.ai
Made by content marketers. Used by better ones.