How to do a newsletter teardown (free scorecard)

We have run more than 14,000 newsletter teardowns through the Newsletrix corpus over the last 18 months, and the pattern in unstructured reviews repeats. Hand the same competitor send to two senior operators. The verdicts agree on the headline. The numeric ratings disagree by 30 to 40%. One cares about subject lines because that is what they ship every week. The other cares about CTA position because that is what they get measured on. Both right, both useless for benchmarking, because the score moves with the reviewer rather than the email.

The fix is a fixed scorecard: seven categories, 32 points, sub-criteria defined in advance. Score blind first, check against historical sends second. This article hands you the rubric, walks one real example through it, and names the mistakes operators make on their first three teardowns.

What a newsletter teardown actually is

A newsletter teardown is a structured audit of one send. Not a SWOT analysis, not an A/B test. A SWOT asks how the publication competes over months. An A/B test asks whether a change beats the control on the next send. A teardown asks how this email, on this Tuesday, stacks up against the rubric.

"I would change the CTA" is not a teardown. It is taste. A teardown has to produce a number you can defend, against criteria you wrote down before you opened the email. The point is reviewer-independence: a teammate who picks up your scorecard six months later should land within two points of where you landed. For the longer version of this distinction, see our newsletter SWOT playbook and the competitor analysis template.

The seven categories every teardown must score

Here is the rubric. Score each category against the sub-criteria, total at the end, write a one-paragraph verdict.

Subject line, 5 points. One point each for length under 50 characters on mobile, specificity (a real noun and verb beats abstraction), curiosity without clickbait, sender-name match, and emoji discipline (zero or one, chosen for inbox-list contrast). See our seven factors of subject-line performance for the long version.

Preheader, 3 points. One point if it is under 90 characters so it renders fully on Gmail desktop and most iOS Mail builds. One if it does not repeat the subject. One if it carries a real payoff line rather than "view in browser" or whitespace padding.

Hook in the first 50 words, 4 points. One point if the first sentence stands on its own without the subject. One if a specific noun appears in the first 15 words. One if the reader knows what they are about to get by word 30. One if the writer earned the open with something the reader did not already know.

Body structure, 5 points. Scan-ability under a five-second skim. Image-to-text ratio at or above 60% text. Section breaks every 180 to 250 words. Body length matched to the publication's stated cadence. A single clear narrative line. One point per criterion; a perfect 5 is rare.

CTA, 5 points. Position above the second fold on a 13-inch laptop. Three or fewer total CTAs (repeated buttons to the same destination count as one). Verb specificity (the verb names the action, not "learn more"). Visual contrast against the body. Repeat-CTA tax: subtract a point if the same destination appears four or more times. Our CTA friction fixes post lists the failure modes we keep seeing.

Deliverability signals, 5 points. SPF aligned. DKIM signing valid. DMARC at quarantine or reject. Image weight under 500 KB. Link domains either match the sender or sit on a clean reputation. Pull these from raw email headers, not from a marketing dashboard. Our SPF, DKIM, DMARC explainer covers the header reads.

Brand consistency, 5 points. From-name matches the sender's public identity. Logo and colour treatment match the homepage. Footer carries a real physical address and a working unsubscribe link. Tone matches the publication's web copy. The reader can identify the brand from the design alone, with the from-name covered.

The 7-step process, 20 minutes per newsletter

The rubric only works inside a process. Skip steps and the score drifts.

Step 1: render the email in Gmail web, Outlook for Windows, and Apple Mail on iOS before you score anything. Mobile-only review misses layout breakage that affects 35 to 45% of B2B audiences who still read on desktop Outlook. Litmus and Email on Acid do this rendering work; a free Gmail and a free Outlook.com account get you 80% of the value if you do not have a paid render tool.

Step 2: score blind. Read the email once, full attention, no scoring tab open. Then score the seven categories from memory. This catches what the email really communicated, before you start hunting for the things the rubric wants you to find.

Step 3: open the scorecard and verify each category against the criteria. Adjust where memory was wrong. Bookkeeping pass.

Step 4: pull the last three sends from the same publication. The teardown number is more useful as a trend than as an absolute. A 24/32 on a publication that scored 28 last month is a different signal than a 24 on a publication that scored 22.

Step 5: write one positive observation per category, even on a low-scoring email. This forces you out of the critique reflex. Every send has at least one decision the writer made well; if you cannot find it, you scored on taste.

Step 6: write the one-paragraph verdict. Name the score, the lowest-scoring category, and the single change you would test next. Three sentences.

Step 7: file the scorecard in a comparable format. Markdown table, Notion database, Sheets row, whichever you use. What matters is that the next teardown reads the same shape. A drawer full of scorecards is a benchmark dataset. A drawer full of Slack messages is not.

Worked example: Morning Brew Tuesday B2B, scored 24/32

We took the Morning Brew B2B Tuesday send from late April through the rubric. The headline number is 24 out of 32. The detail matters more than the headline.

Subject: 4/5. Strong specificity, proper noun in the first 30 characters, sender-name match. The held-back point is emoji discipline: a single decorative emoji that does not earn its inbox-list real estate. Preheader: 1/3. Length 119 characters; iOS Mail clipped at 67 on the test render, cutting the payoff line mid-clause. Mechanical fix; the editorial team almost certainly knows the constraint.

Hook: 4/4. First sentence works without the subject, a real noun lands by word 11, the reader knows the shape of the issue by word 28. This is what Morning Brew is famous for. Body structure: 4/5. Scan-ability is excellent, image-to-text ratio above 60% text, section breaks land on rhythm. The dropped point is body length, about 18% longer than their stated five-minute promise.

CTA: 2/5. The failure. Five buttons to four destinations, with the primary CTA stacked past the third fold on a 13-inch laptop render. Position is the problem. Deliverability: 4/5. Full alignment on SPF, DKIM, DMARC at reject, image weight a hair over 500 KB. Brand consistency: 5/5.

Two losses on a 24/32: preheader cutoff and CTA stacking. Both fixable inside one send cycle. If we were running their A/B test queue, the preheader fix ships first because the cost is zero. The CTA test takes longer to set up cleanly. See reverse-engineering a competitor newsletter for the longer pattern.

Common teardown mistakes

Three failure modes show up across the first hundred or so teardowns most operators run.

Scoring on taste rather than rubric. Dislike the typeface, body structure score drops. Like the tone, hook score goes up. The moves are invisible to the reviewer doing them. If you catch yourself adjusting one category after reading another, you are scoring on taste.

Ignoring render differences across clients. Gmail Promotions, Apple Mail, and Outlook 2019 render the same HTML differently. A CTA above the fold in Gmail web sits below it in Outlook desktop. Score from the worst render, not the best. See how Newsletrix compares to Litmus for the longer render-testing breakdown.

Not comparing the score to the sender's own previous sends. A 24/32 is decent. A 24 on a publication that averaged 29 last quarter is a regression. Without the trend, you are scoring against your assumptions about what a good newsletter looks like.

The free 32-point scorecard, copy-paste

Here is the markdown table. Paste it into Notion, Sheets, or a plain text file and use it as the template for every teardown.

| Category              | Max | Score | Notes |
|-----------------------|-----|-------|-------|
| Subject line          | 5   |       |       |
| Preheader             | 3   |       |       |
| Hook (first 50 words) | 4   |       |       |
| Body structure        | 5   |       |       |
| CTA                   | 5   |       |       |
| Deliverability        | 5   |       |       |
| Brand consistency     | 5   |       |       |
| TOTAL                 | 32  |       |       |

Verdict (3 sentences max):
- Headline number:
- Lowest-scoring category:
- Single change to test next:

Score the categories independently. Total at the end. Verdict last. Resist the urge to add an eighth category for the thing you noticed that the rubric does not capture; either fold it into an existing category or write a separate note. The rubric only stays useful if it stays fixed.

When to automate the teardown

Five competitors monthly is fine by hand. The scorecard takes 20 minutes, five of them once a month, an hour and forty minutes for the whole exercise. Cheap, comparable, useful.

Ten or more competitors weekly is where the manual process breaks. We have watched marketing teams try. The scorecard gets shorter, the verdict gets shorter, and within three months the teardown becomes a Slack screenshot with a thumbs-up. The benchmark dataset stops getting built.

At that scale, the rubric stays the same and the work goes automated. Newsletrix runs the same seven categories across every send we ingest, scores deliverability from the raw headers, surfaces the trend over the last 12 sends, and writes the one-paragraph verdict for you to edit. The tradeoff is honest: you lose the muscle memory you get from doing teardowns by hand. First month on automation, teardown your top three competitors manually anyway and check your score against ours. Land within two points and automation is doing the job. Land further off and the rubric needs tuning, not the tool.

For the data-collection side, see tracking competitor newsletters. The teardown is what you do with the collection.

Frequently asked questions

What is a newsletter teardown?

A structured audit of one newsletter send, scored against a fixed rubric. The point is to produce a comparable number for subject, preheader, hook, body, CTA, deliverability, and brand consistency, so two reviewers reach roughly the same conclusion. Without a rubric, the same send scores 30 to 40% differently between two operators.

How long does an email teardown take?

Around 20 minutes once you have the scorecard in front of you. Five minutes to render in Gmail, Outlook, and Apple Mail. Ten minutes for the seven categories. Five minutes to write the verdict and pick the single change you would test next. Faster than 15 minutes usually means you skipped the render check.

What should a teardown scorecard include?

Seven categories on a fixed point total: subject line (5), preheader (3), hook in the first 50 words (4), body structure (5), CTA (5), deliverability signals (5), brand consistency (5). Total 32 points. Each category needs explicit sub-criteria so a different reviewer reading the same email lands within two points of you.

How do I teardown a competitor newsletter ethically?

Subscribe with a real address you own, score what they sent, and keep your notes private to your team. Do not republish their copy, do not screenshot the body for public posts, and do not run their content through scrapers that ignore robots.txt. The point is to read the pattern, not to repost the artefact.

What is the difference between an email teardown and an A/B test?

A teardown is a scoring exercise on what was sent. An A/B test is a controlled experiment on a future send. The teardown tells you which category is underperforming so you know what to test. The A/B test tells you whether the change you made beats the control. Run the teardown first, then design the test against the lowest-scoring category.

Related reading