Newsletters: the feed and the detail view

What this screen is

Newsletters is the index of your collection. Every email that reached your tracking address and finished its AI analysis appears here as one row, newest first, with the brand, the subject, the market segment and two estimated engagement rates. Every other screen in Newsletrix is an aggregation of these rows. Come back here when the question is about one specific email rather than a pattern.

The table itself is deliberately thin. Click any row and a near-fullscreen detail view opens with seven panels. You get the original email rendered pixel for pixel, the full written evaluation, each estimated metric with the reasoning printed under it, an authorship fingerprint, and a prediction of what Gmail, Apple Mail, Outlook, ChatGPT and Perplexity would say about the email if asked to summarise it.

There are two tabs. Newsletters is the sortable, searchable table. Previews is the same collection rendered as a wall of live thumbnails, for the times you recognise an email by its layout faster than by its subject line. Both tabs open the same detail view.

This screen does not compare, rank or trend. You will not find a brand scorecard here, a week-over-week movement figure or a cross-brand chart. Emails that arrived but have not been analysed yet are missing too, so the total in the table header counts finished analyses rather than received mail.

Unlike the aggregate screens, this one needs no minimum sample. One analysed newsletter is enough for it to be useful. The Brand and Email type dropdowns stay empty until at least one analysis exists, because their options come from the emails you have already had analysed.

Open it when

• A competitor just sent something and you want the email and the AI verdict side by side.
• You need to find one specific send again: the Black Friday email from a given brand, last November.
• A layout is worth stealing and you want to see how it renders on desktop versus mobile.
• You want to know whether an AI inbox summary would keep or kill that email's call to action.
• A client or colleague with no login needs to see one competitor teardown.
• You are eyeballing a stack of recent sends visually rather than reading subject lines.

Screen anatomy

The Newsletters tab with the date range revealed and applied. The numbers match the list underneath.

1. Tab strip Two tabs: Newsletters (the table) and Previews (the thumbnail wall). Newsletters is the default. Each tab keeps its own set of filters, and the Previews tab has fewer of them.
2. Filter bar Brand is a multi-select, so you can hold several brands at once. Email type is a single choice. Search runs across both the subject line and the body text. The date range lives behind the More filters toggle, which reads Hide filters once it is open, and the small badge on it counts how many of the two date fields are filled. Clear all only appears once at least one filter is set, and resets everything in one click.
3. Newsletters table Seven columns: Date, Brand, Subject, Segment, Est. open rate, Est. CTR and Actions. The count in the card title reflects the active filters, not the size of your whole archive, and counts only newsletters with a completed analysis. Only Date, Brand and Subject are sortable; the default is Date, newest first. The whole row is clickable.
4. Pagination footer Twenty rows per page, fixed. There is no page-size control. The footer appears only when there is more than one page.
5. Section header and help button The title, the subtitle, and a question-mark button that opens a short description dialog. That dialog carries the same wording on every screen, and it mentions bulk actions, archive organisation and export options this screen does not have. Trust this page over that dialog.
6. Est. open rate and Est. CTR columns Both are estimates for that single email, printed to one decimal place. A dash means the analysis produced no usable value.
7. Actions column The View analysis button opens exactly the same detail view as clicking anywhere else on the row. It is there to make the row target obvious.

Every panel explained

The newsletters table

One row per analysed newsletter, in reverse chronological order by the date in the email's own header.

How it is calculated. The list covers your own account. Select a workspace in the workspace switcher and it narrows again, to the tracking addresses tagged with that workspace. Whatever filters you set narrow it further still. Brand takes one or more values, email type takes one, search matches the subject and the body text, and the date range applies to the date in the email header. Twenty rows to a page. Only Date, Brand and Subject can be sorted. Clicking one of those headers sorts ascending, and clicking it again flips to descending. The arrow icon appears only on the column you sorted by. Both percentage cells are rounded to one decimal and carry a percent sign.

How to read it. Treat it as a chronological archive with a search box. Nothing in it is ranked. Est. open rate and Est. CTR are informed guesses about one email, drawn from its content and from published benchmarks, so they will tell you which of a brand's sends looks strongest and nothing whatever about what the sender really measured.

The Previews tab

A four-across card grid. Each card carries the subject line, a purple brand pill, a blue email type pill, a live scrollable rendering of the original email, the capture date, an optional estimated open rate and a View details button.

How it is calculated. The same account and workspace narrowing as the table, and the same brand and date filters, but no search, no email type filter and no sort control. Order is fixed: newest capture first. Each card shows the email itself, and a tall one scrolls inside the card.

How to read it. Use it as a visual scan of recent captures. It is faster than reading subject lines when you are looking for a particular creative treatment, a seasonal campaign look, or a layout you half remember.

The detail view

Clicking a row opens a near-fullscreen dialog with a fixed left rail of seven panels and a scrollable content pane. The rail always resets to Email preview when the dialog opens, and its footer repeats the quick facts: brand, date and email type.

Detail view: Email preview

Five sub-tabs: Desktop, Mobile, Screenshot, Source and Plain text. Desktop wraps the email's own HTML in a browser frame whose fake address bar shows the brand and the first 60 characters of the subject. Mobile wraps it in a 380 pixel phone frame with a mail app header.

How it is calculated. Desktop and Mobile render the email live in your own browser. Screenshot is a picture taken in advance at fixed sizes, 1280 by 800 for desktop and 375 by 812 for mobile. Source shows the sender's raw HTML with line numbers, and Plain text shows the text part the sender included.

How to read it. Desktop and Mobile tell you how your browser renders the email, which is close enough for layout ideas. Screenshot is the reference you can reproduce, since it was taken at a fixed size rather than at whatever width your window happens to be.

Detail view: AI Inbox Preview

A header showing an overall score out of 100 labelled AI readiness and a one-sentence verdict, then up to five cards: Gmail AI (Gemini summary), Apple Intelligence (Mail summary), Outlook Copilot (Key points), ChatGPT agent (Intent extraction) and Perplexity agent (Answer extraction), always in that order.

How it is calculated. For each of the five assistants, Newsletrix predicts the summary that assistant would show, one to three sentences and roughly 600 characters. Alongside it come a format style, a predicted priority, the key points that survive and the ones that get dropped (up to eight of each), a verdict on whether the call to action survives, a verdict on whether the brand is named, a visibility score, a risk note and an optimisation tip. Both the overall readiness score and each visibility score are whole numbers from 0 to 100, and the colour bands are the same for both: 70 and above is green, 40 to 69 amber, below 40 red.

How to read it. The overall score answers one question: will this email survive being summarised in the inbox. Per card, read the predicted summary first, then the survives and dropped columns, then the two badges that tell you whether the action and the sender's name make it into the one or two lines a recipient actually sees.

Detail view: Basic info

The identity of the send: subject, brand, date, email type and segment. Under that sits a communication style card (tone, formality, emotional appeal, personalization, language style, voice characteristics, key messages), then the benefits the email offers and the content categories it promotes.

How it is calculated. Email type is one of promo, educational, news, abandoned cart, transactional or other. Each benefit carries a prominence of high, medium or low, shown as a green, amber or grey badge.

How to read it. This is the who and the how of the email: which audience slice it targets, what tone it uses, and what it promises.

Detail view: Performance

The longest panel. It opens with a written evaluation, then four estimate tiles, then strengths and weaknesses side by side, the audience and growth strategies Newsletrix reverse-engineered from the email, a design and length assessment, and a compliance and subscribe link card.

How it is calculated. The written evaluation runs 200 to 400 words in an analyst's register, covering content quality, technical execution, informational value and two or three high-impact improvements. Under it sit four tiles: Est. open rate, Est. CTR, Images vs text and Confidence. The first three stay inside 0-100 and are shown to one decimal. Images vs text is the share of visual content, so 70.0 means 70% images and 30% text. Confidence runs on a 0-1 scale rather than as a percentage. On the compliance card you get a Compliant or Has issues badge, a 0-100 score, the required elements found and the issues detected. The subscribe link is the one item on the card that is not an estimate. It is a real link out of the email, and the card reads Not found when none was picked up.

How to read it. Read the written evaluation before the numbers. Then read each tile with the note underneath it. A note that says estimate or assumption is telling you the figure was a guess, which is your signal for how much weight it deserves.

Detail view: AI Fingerprint

A verdict badge, four score rings (AI Likelihood, Human Editing, Originality, AI Preview), a pattern analysis paragraph, and a list of the individual signals that shaped the scores, each badged AI-leaning, Human-leaning or Neutral.

How it is calculated. The four scores are whole numbers from 0 to 100. The verdict badge is derived from AI Likelihood alone: 70 or above reads Likely AI-generated, 40 to 69 reads AI-assisted, below 40 reads Predominantly human. Ring colour is green at 67 and above, amber from 34 to 66 and red below 34. On AI Likelihood the ring is inverted, so the band is read against 100 minus the score and a high number comes out red. The other three rings take the score as it stands. A missing score shows a grey ring reading n/a. AI Preview is one blended score covering both inbox summary readiness and citability by generative search engines.

How to read it. High AI Likelihood with low Human Editing means the copy reads machine-written and unrevised. The pattern analysis paragraph tells you which signals produced that read, which is more useful than the score itself.

Detail view: Recommendations and Next actions

Two playbook panels. Recommendations lists up to five concrete changes to this email, each with an optional rationale and an expected impact line. Next actions lists three to five imperative moves with badges for Timing, Channel, Offer type and Expected impact, plus a market rationale and the competitor signal that triggered it.

How it is calculated. Recommendations are capped at five, and the expected impact carries a percentage wherever one can be estimated. Next actions come from what this competitor just did, weighed against the season, any upcoming events, the current opportunity window and the angles trending in your market. Channel is one of email, email and social, email and paid, or all.

How to read it. Recommendations answer "what is wrong with this email". Next actions answer "what should we do next, given this email and the calendar". They are different questions and the two lists rarely overlap.

Share link and the public page

The share button in the detail toolbar opens a small dialog holding a read-only link and a copy button. The link opens a standalone public page with a branded header, the subject and a seven-tab teardown covering Basic info, AI summary, Performance, Content, Marketing, Recommendations and Email preview.

How it is calculated. The link is the app address plus the newsletter's own id. The public page carries the same analysis as the detail view, laid out differently. All four performance tiles are always drawn there, with N/A printed where the detail view would hide one. A Marketing tab appears that the detail view does not have, covering call-to-action phrases, CTA destinations, promotional products, discount patterns, seasonal campaigns and pricing communication. The email's sections, the required elements found and the issues detected are all listed, while the compliance badge, the compliance score and the subscribe link are dropped. Recommendations gain improvement priorities with an expected impact figure. The email preview offers only Desktop and Mobile.

How to read it. It is the shareable version of a teardown, suitable for a client who will never log in. Next actions is the one panel it does not carry.

Metric definitions

Filters and controls

Some things are deliberately absent, so stop looking for them. There is no spreadsheet or CSV export of the list, no bulk selection, no folders or archive organisation, and no page-size control. The only export is the per-newsletter PDF inside the detail view. The Previews tab has no search, no email type filter and no sort. And there is no control anywhere to revoke a share link once it has been created. The help dialog behind the question-mark button lists several of these as if they exist. It carries the same wording on every screen and is wrong on this one.

How to use it

1. Narrow before you scroll Pick the brand or brands, then open More filters and set the date range to the period you are studying. Use search only when you remember an actual phrase; it matches the body text as well as the subject, so it is broader than it looks.
2. Open the send, not the row summary Click anywhere on the row. The two estimated rates in the table are a hint, not the answer; the reasoning behind them is one click away.
3. Look at the email before you read about it Start on Desktop, switch to Mobile, and use Screenshot when you need a fixed-viewport reference you can put in a deck. Form your own view of the creative before the analysis colours it.
4. Read Performance top down The written evaluation first, then the four tiles with the notes underneath them, then strengths and weaknesses. The notes tell you which numbers were reasoned through and which were assumed.
5. Check whether the offer survives the inbox Open AI Inbox Preview and look at the CTA and brand badges across the five assistants. If a competitor's discount vanishes from every predicted summary, that is a mistake you can avoid in your own next send.
6. Turn it into work Recommendations for fixing an email like this one, Next actions for what to do this week. Share the link when a client needs to see the teardown without a login.

Limits and caveats

• Every percentage on this screen except the filter counts is an AI estimate, not a measurement. Newsletrix receives these emails as an ordinary subscriber and can never see a sender's real opens or clicks.
• Only newsletters with a completed analysis appear. An email that arrived but has not been analysed is invisible here, in both tabs and in the header count.
• Newsletrix is tuned for the e-commerce footwear vertical. The segment wording, the category vocabulary and some of the benchmark reasoning lean that way, which matters if you track brands in a different market.
• A figure that would land outside its range is shown at the nearest end of that range instead, so a weak guess can arrive looking like a clean 0 or 100.
• Zero is displayed inconsistently: an Est. CTR of exactly zero shows as a dash in the table while an Est. open rate of zero shows as 0.0%. Open the detail view to tell an absent value from a genuine zero.
• The email previews render the sender's original HTML with remote images intact, so tracking pixels in an email body can fire when you view a preview.
• Nothing in the lists of strengths, weaknesses, recommendations, issues and key messages is ranked. A newly analysed newsletter can take up to two minutes to show in the list, and a brand or email type you have never had before can take up to fifteen minutes to show in the filter dropdowns.

FAQ

Related screens