Why SaaS Emails Score 3.4 Out of 10

The number comes from 59 real teardowns published over 18 months. Companies audited include Notion, Revolut, dbt Labs, Perplexity, Linear, and 54 others across SaaS, fintech, AI tools, and enterprise software. Each email was scored on 7 structural criteria. The average original score is 3.4 out of 10. The average rebuilt score is 9.0.

That gap, 5.6 points, comes from fixing structural decisions. Not words. Decisions about what information appears first, where the consequence sits relative to context, and whether the reader is addressed as a guest or an owner.

Three patterns appear in over 80% of audited emails. They account for the gap between 3.4 and 9.0.

Pattern 1: Guest Language CTA

A Guest Language CTA uses a verb that places the reader outside the action. "Learn more." "See features." "Explore now." "View the update." The reader is positioned as a visitor considering whether to engage, not a person already deciding to act.

The problem is not politeness. The problem is that guest language removes ownership. When you write "Learn more," you are naming what your brand will show, not what the reader will do. There is no action the reader can identify as their own. The CTA asks for nothing specific and gets nothing in return.

The repair: identify the exact decision the reader is being asked to make, then write the CTA as if the reader is already making it. "Fix my reporting" names a problem the reader owns and resolves. "Learn more" names nothing the reader is doing.

Pattern 2: Feature-First Bias

Feature-First Bias is an opening structure that leads with what the product does before the reader has any reason to care. The email announces a feature before establishing the consequence of that feature for the reader's situation.

The structural failure is sequencing. The reader has not yet been given a problem to solve. The feature arrives before the context that would make it relevant. Because the reader has no active problem in mind, the feature registers as noise and the email is closed.

The repair requires one structural inversion: state the reader's problem or gain before naming the feature. This is not a rewrite. It is a reorder. The same sentences work; the sequence is wrong.

Pattern 3: Filing Label Subject

A Filing Label Subject announces the category of the email rather than a consequence for the reader. "New Feature: Advanced Reporting Dashboard." "Product Update: June 2026." "Introducing AI Inbox Sorting."

The brain processes these as metadata. The reader routes the email to a mental folder labelled "read later." That folder is rarely opened. The filing label subject does not create a gap between the reader's current state and a desirable outcome. It only identifies what type of email this is.

The repair is consequence-first subject construction. Name what the reader gains or avoids, not what the company has shipped. The feature can appear in the preview text or the body. The subject line carries only one job: give the reader a reason to open.

Why these are architecture problems, not copy problems

All three patterns persist because they feel safe. Guest language is inoffensive. Feature-first structure mirrors how product teams think about their work. Filing labels are factually accurate. None of them are errors in the traditional copywriting sense.

They are errors in sequencing. In the order of information. In who holds ownership of the action. These decisions are made before the first word is chosen, in the architecture of the email, not in the editing of its sentences.

Editing a Guest Language CTA into ownership language is a one-sentence change. Editing a Feature-First opening into a consequence-first opening is a structural inversion that takes under 5 minutes. Filing Label subjects become consequence signals with a single rewrite of 10 words or fewer.

The 5.6-point gap between 3.4 and 9.0 is not the result of extensive rewriting. It is the result of three structural corrections applied before the email is sent.