- Where
- Background, no sidebar entry
- App route
- None, applies to every screen
- Answers
- Where every number on a chart comes from
- Needs
- One tracking address, one delivered newsletter
- Updates
- New mail picked up every couple of hours
What this screen is
Every other article here maps a screen. This one has none. It follows a single newsletter from the moment it reaches your tracking address to the moment a tile on Overview changes, and covers the furniture around every screen, from the sidebar and its workspace selector down to the plan limits that decide how much gets analysed.
What arrives sets the ceiling on what the app can know. Newsletrix receives a newsletter the way any subscriber does, with the subject, the sender details, the designed and plain-text versions of the message, its links and its send time. It never sees the sender's list size, opens, clicks or revenue. Anything labelled est. is an estimate Newsletrix made from the content. Everything else is a plain count or a fixed formula applied to the email.
Two kinds of number come out of a newsletter. The AI reading produces about 60 details, from brand, prices, section order and CTA destinations through tone, compliance, strengths, weaknesses and recommendations to an AI-authorship fingerprint and a summary of 200 to 400 words. The rest come from fixed formulas, among them a spam score, a promotion percentage, a 0-100 technical score with an A to F grade, word and link counts, sentiment polarity and two readability indices. Most screens show the two kinds side by side, and the metric table below marks which is which.
Open it when
- You just signed up and want to know where the tracking address comes from and what to do with it.
- A newsletter arrived hours ago and is not on the dashboard yet.
- You want the full navigation list in one place to find a screen.
- You are choosing a plan and want the real weekly and address limits, not the pricing-page summary.
- A colleague asks where the open rate number comes from.
- A number has not moved since a new newsletter was analysed.
Screen anatomy
The first figure is the app shell that wraps every screen, shown around Overview with sample data. The second is the Get Started card that follows your newest newsletter while it is being analysed. Marker numbers continue across both.
Overview
Hi arzzen 👋 - here's a summary of your newsletters.
Welcome to Newsletrix
- Workspace selector Scopes every chart to newsletters received at addresses carrying the chosen label; All workspaces means no filter. Choosing one reloads the app. The purple Create workspace entry opens the Workspaces page.
- Total newsletters tile Newsletters with a finished analysis, with a sparkline and a trend chip. Newsletters skipped for quota are not counted.
- Avg open rate (est.) tile Average of the per-newsletter estimate wherever one exists, rounded to two decimals and shown to one. Never measured from a real send; the line under it sets yours beside a peer figure.
- Top brand tile The brand named most often, its email count, and Share: that count divided by total newsletters, rounded to a whole percent. Top send day does the same with the weekday and calls it Concentration. Both read N/A with no data.
- Search row Opens the command palette (⌘K or Ctrl+K toggles, Esc closes). It lists 21 of the 23 destinations plus an Actions group; Next actions and Blacklist are missing, and three labels differ from the sidebar: Send time distribution, Brand comparison, Get started.
- Trial strip and user block The yellow strip appears only on invitation-code trials, counting down from the code's length (30 days by default) to period ended. Below it: initials, username and tier name. Names exist for Free, Basic, Standard, Premium and Unlimited; Founders reads Unknown Tier. The menu holds Profile & Settings, About & FAQ, the theme switch and Log out.
- Tracking address card Addresses are news+ eight random lowercase letters or digits @newsletr.email. The first is generated when your account is activated; more can be added here or in Settings, each with an optional workspace label.
- Progress strip Pending, Fetching, AI analysis, Done. The card updates itself every 8 seconds until an analysis lands. Done means an analysis is there, AI analysis means the newest newsletter has arrived without one, Pending means nothing has arrived at all. In practice the strip jumps straight from Pending to AI analysis, and Fetching is rarely on screen long enough to see.
- Latest newsletter row An amber In analyzer queue chip and N in progress until a result exists, then an emerald Analyzed chip. There is no skipped state: a newsletter held back by the weekly limit keeps the amber chip indefinitely.
- Navigation groups (no marker) 23 destinations in four groups. Get Started, Overview, Recommendations and Next actions sit at the top, then Analysis (Newsletters, Brands), Content (Sentiment, Subject lines, Promotions, Campaigns, CTA elements, Structures) and Insights (SWOT analysis, Readability, DateTime distribution, Trending topics, Link positions, Sections, Duplicates, Google annotations, Sending platform, Subscribe links, Blacklist). All are visible on every plan; only an exact URL match highlights an item. Compare, Overview reports, Editorial cadence, Workspaces, Settings and About have no entry and are reached from other screens or by URL.
- Shell states (no marker) A Loading… spinner while the session is checked, a blocking Subscription expired modal on a lapsed plan (everything except your own details and logging out stops working), an impersonation banner when an administrator views your account, and an off-canvas rail on narrow screens. The rail-edge button collapses the sidebar from 256px to 64px; administrators also get an Admin panel dropdown.
Every panel explained
None of this has a screen of its own. The headings below follow a newsletter from your tracking address to a finished analysis, then cover the two shell elements that change what every other screen shows.
Tracking addresses
Where all data enters. The address tells the app which account a newsletter belongs to.
How it is calculated. The first address is created on activation: after email confirmation on Free, after payment on paid plans. Each new one is checked against your active count: Free 1, Basic 1, Standard 5, Premium 15, Unlimited uncapped; over the limit you see Email address limit reached. Deactivating an address keeps what it already collected, still counted in its workspace.
How to read it. One address per source you want to keep apart; its workspace label is what the sidebar selector filters on later.
Collection, every two hours
How new mail reaches Newsletrix.
How it is calculated. New mail addressed to any of your active tracking addresses is picked up every couple of hours. Nothing needs forwarding, and nothing you do makes a pickup happen sooner.
How to read it. Worst case, a newsletter waits just under two hours to be picked up. Other parts of the app keep their own clocks. Recommendation examples and subscribe-link confirmations refresh through the day, the weekly digest arrives on Monday morning, and the Blacklist screen carries the date it was last checked.
The weekly limit
What decides whether a collected email is analysed at all.
How it is calculated. A newsletter has to be addressed to one of your tracking addresses. Anything else is dropped, because it belongs to no account. Your weekly allowance is 2 on Free, 5 on Basic, 15 on Standard, 50 on Premium and Founders, and uncapped on the internal Unlimited tier. Over the limit, a newsletter still turns up in your list but never gets an analysis. Weeks start on Monday, and the counter rises by one per finished analysis, never falls and does not roll over. A sending platform Newsletrix cannot identify is shown as unknown, with no confidence figure beside it.
How to read it. Settings colours usage green below 80% of the limit, yellow from 80%, red at 100%.
An email skipped for quota and one that is still waiting look identical. Both wear the amber In analyzer queue chip. When it will not clear, check Used this week first.
The AI reading
Where most of the detail behind the other screens comes from, including the inbox assistant previews.
How it is calculated. Every newsletter is read against the same checklist of 16 analysis sections and about 60 details, against the same market context that drives Next actions (season, active campaigns, upcoming events, opportunity window, trending angles). Estimated values (open rate, CTR, click probability, image-to-text ratio, word count) are labelled as estimates, with the open rate, the CTR and the image ratio each carrying a note on their basis, and every analysis includes a confidence estimate from 0 (a guess) to 1 (high) plus notes on missing data. Five inbox assistants get a preview each, for Gmail Gemini, Apple Intelligence, Outlook Copilot, a ChatGPT agent and a Perplexity agent. Each preview carries a 1 to 3 sentence summary under about 600 characters, a format style (one-liner, bullets, paragraph, priority flag), a predicted priority (high, normal, low, promotions), up to 8 key points surfaced and 8 lost, whether the CTA survives, whether the brand is mentioned, a 0-100 visibility score, a risk note and a tip. Above the five sits an overall 0-100 inbox readiness score and a verdict.
How to read it. Ask for the same newsletter to be read again and the answer can differ; the newest result replaces the one before it. The inbox previews are predictions of how each assistant would treat the email, never a record of what one actually did, and a preview can be older than the analysis beside it.
Blank fields and re-analysis
Why a field on a newsletter can be empty, and what changes when the same email is read again.
How it is calculated. A value shows up only when the reading produced one that makes sense for the field it belongs to. Anything else is left blank rather than guessed at, and a reading that cannot be used at all leaves the newsletter with an empty analysis. Read the same newsletter again and the new result replaces the old one whole, lists included, so nothing from the earlier reading survives underneath. Alongside the full analysis text, each newsletter carries 3 to 5 next actions (action, timing, channel, offer type, market rationale, competitor signal, expected impact).
How to read it. A blank value means the analysis produced nothing usable for that field. It does not mean the email lacked it.
Fixed formulas and content features
Everything that is computed rather than estimated.
How it is calculated. The spam score adds points for SPF, DKIM and DMARC failures and for spam headers the receiving system attached, capped at 10. Promotion percentage adds capped contributions from discount patterns in the text, from promotional keywords and from the discounts the reading found, capped again at 100. The overall score starts at 100, subtracts for spam, for size above 100 KB and for promotion above 70%, adds a little when the open-rate estimate exceeds 20%, and is graded A to F; exact increments are in the table below. Content features are plain counts and indices - word count, link count, sentiment polarity of subject and body, Flesch reading ease and Flesch-Kincaid grade. A subscribe link and Gmail promotion markup are picked out of the email when they are there. Every finished analysis adds one to Used this week, and the first one on an account sends you a one-time confirmation email.
How to read it. These numbers are reproducible; the same email always yields the same spam score and grade. A blank annotation or subscribe link means none was found.
BIMI is recorded but never scored, temporary or permanent authentication errors add nothing, and an SPF softfail is currently scored like a fail.
The workspace selector, and why a number lags
The one global filter, and the gap between a finished analysis and the screen in front of you.
How it is calculated. A workspace is the optional label on a tracking address. Choosing one reloads the app and limits every screen to newsletters received at addresses carrying that label. Names allow letters, digits, spaces, underscores, hyphens and dots, up to 100 characters, and a new workspace shows up in the selector at once. Figures move slower than the selector does. A newsletter turns up in the lists and tables within about 2 minutes of being analysed, while the Overview tiles and charts can trail it by up to 10 minutes.
How to read it. All workspaces means no filter, and deactivated addresses stay included. A fresh analysis can take up to 10 minutes to move an Overview tile even though the newsletter itself is already listed. Reloading the page shortens that wait but does not end it.
Metric definitions
| Metric | What it means | How it is produced | Scale |
|---|---|---|---|
| Total newsletters | Newsletters with a finished analysis | Collected newsletters that have an analysis, within the active workspace, brand and date filters | Whole number |
| Total brands | Distinct brands Newsletrix identified | Distinct brand names across analysed newsletters; the brand filter is ignored on purpose | Whole number |
| Avg open rate (est.) | Mean of the estimated open rate | Averaged over newsletters that have an estimate, rounded to 2 decimals; 0% when there is no data | 0-100%, estimate |
| Avg CTR (est.) | Mean of the estimated click-through rate | Averaged over the same newsletters as the open rate, so a CTR estimate without an open-rate estimate is excluded | 0-100%, estimate |
| Top brand / Share | Most frequent brand and its share | Brand with the most analysed newsletters; Share is its count divided by total newsletters, rounded to a whole percent | 0-100% |
| Top send day / Concentration | Weekday with the most sends | Weekday of each newsletter's date header; Concentration is the top day's count divided by the total, rounded | Monday-Sunday, 0-100% |
| Weekly analysis limit | Analyses your plan allows per week | Free 2, Basic 5, Standard 15, Premium 50, Founders 50, Unlimited none; the week starts on Monday | 2 / 5 / 15 / 50 / unlimited |
| Used this week | Analyses consumed since Monday | Plus one after each successful analysis; never decrements, no rollover. Green below 80% of the limit, yellow from 80%, red at 100% | Whole number |
| Tracking address limit | Active addresses your plan allows | Free 1, Basic 1, Standard 5, Premium 15, Unlimited none; counted against active addresses only | 1 / 5 / 15 / unlimited |
| Spam score | Header-based deliverability risk | SPF fail +3 (softfail scores the same), neutral +1; DKIM fail +3, none +2, neutral +1; DMARC fail +3, none +1; plus the whole-number value of any spam-score header; +5 when a spam-status header says yes; capped at 10 | 0-10, lower is better |
| Promotion percentage | How promotional the body reads | Discount matches x 15 (max 40) + distinct promo keywords x 3 (max 30) + discount patterns from the reading x 10 (max 30); capped at 100; 0 on an empty body | 0-100% |
| Overall email score | Composite technical quality with a grade | 100, minus 5 per spam point, minus 1 per 10 KB above 100 KB (max 20), minus a third of promotion points above 70, plus a fifth of open-rate points above 20 when an estimate exists; clamped 0-100. A 90+, B 80+, C 70+, D 60+, else F | 0-100 and A-F |
| Confidence estimate | How much to trust this particular reading | Set during the reading, on every newsletter | 0 (guess) to 1 (high) |
| AI content fingerprint | AI likelihood, human editing, originality, AI preview optimization | Four scores set during the reading, plus a short verdict and signals each leaning ai, human or neutral | 0-100 each |
| Discount depth bucket | Depth of a discount quoted in the email | A range such as 20-40% becomes its midpoint; BOGO, 2 for 1, half price and half off count as 50; otherwise the first N%. A depth outside 0 to 100% gets no bucket. Under 10 is 0-10%, under 25 is 10-25%, under 50 is 25-50%, else 50%+ | 4 buckets |
| Price war week | A week when several brands discounted hard | A Monday-start week in which at least 2 distinct brands each sent a newsletter with a maximum depth of 25% or more; the 12 most recent are shown | Brands per week |
| Tone bucket | Coarse tone grouping | From emotional appeal and communication tone, in order: urgency gives Urgent, and only the emotional appeal can produce it; excitement or playful gives Excited; comfort, trust, aspiration, belonging, professional, formal or authoritative gives Calm; casual, friendly or conversational gives Friendly; else Neutral | 5 buckets |
Filters and controls
| Control | What it does | Default |
|---|---|---|
| Workspace selector | Scopes every screen to newsletters received at addresses carrying that label, then reloads the app | All workspaces |
| Brand filter | On analytics screens; accepts one or more brands. The Total brands tile ignores it | All brands |
| Date range | From and to dates compared with each newsletter's date header. On Overview the inputs are switched off, so it shows all time | All time |
| Sidebar collapse | Shrinks the 256px rail to a 64px icon rail with tooltips; hidden on mobile | Expanded |
| Search / command palette | Jump to any of 21 screens or run an action; ⌘K or Ctrl+K toggles it, Esc closes | Closed |
| Refresh (Get Started) | Checks for progress now instead of waiting for the next 8-second update. It does not start an analysis | Updates every 8 seconds |
| Change subscription tier | In Settings; Basic, Standard, Premium and Founders can be chosen, Unlimited cannot. Cancel subscription sits in the same card | Current plan |
| Next actions window | Lookback in days and a cap on actions, clamped to 1-365 days and 1-200 actions; non-numeric input falls back to the default | 30 days, 50 actions |
Three things are deliberately missing. There is no global export; exports live inside individual screens such as Overview reports. No per-newsletter progress or skipped state exists either. And no button anywhere makes Newsletrix collect mail early; Refresh only shows what has already happened.
How to use it
- Create the address and subscribe Copy your news+ address from Get Started, subscribe it to three to five competitors, and label it if you will ever need to separate clients or markets.
- Wait for the next pickup Mail is picked up every couple of hours, so the wait is under two hours. Watch the Your first newsletter card move from Pending to AI analysis to Done. The first result also sends you a one-time confirmation email.
- Check the quota before you judge a gap In Settings, compare Used this week with your weekly limit. An amber In analyzer queue chip next to a full counter means skipped, not stuck.
- Open the newsletter itself Read the confidence estimate, the notes on missing data and the note behind each est. figure before trusting a single number.
- Give a stubborn number time Overview can sit up to 10 minutes behind a fresh analysis. Newsletter lists lag about 2 minutes.
Count the newsletters your tracked senders ship in a typical week and keep it below your plan's weekly limit. Anything over still turns up in your list but never gets an analysis, and the app will not tell you which ones were dropped.
Limits and caveats
- Open rate, CTR, click probability, image-to-text ratio and word count are estimates, not measurements. The confidence estimate (0 to 1) says how much to trust the reading behind them.
- The recommendations section leans on footwear e-commerce and names one shoe brand, so wording from that trade can surface in other industries.
- The AI reading is not repeatable. Analyse the same newsletter again and the numbers can change; the newest result overwrites the one before it, lists included.
- A newsletter analysed before Newsletrix started producing next actions comes back empty on that screen.
- An unusable reading leaves an empty analysis behind, and a value that could not be used is simply blank, never flagged as missing.
- Founders accounts currently get 1 tracking address although the plan card promises 15, appear as Unknown Tier in the sidebar, and see the Basic plan card in Settings. The weekly limit of 50 is applied correctly. Contact support if you are on Founders and need more addresses.
FAQ
I forwarded a newsletter and nothing appeared. Why?
Forwarding is not supported. Newsletrix collects only mail addressed to one of your tracking addresses, and it uses that address to work out whose account the newsletter belongs to. A forwarded message carries your own address, matches nobody and is skipped. Subscribe with the tracking address instead.
How often does the analysis run?
New mail is picked up every couple of hours, so a newsletter waits at most just under two hours, then seconds more to be read and scored. Recommendation examples and subscribe-link confirmations refresh through the day, the weekly digest arrives on Monday morning, and the Blacklist screen carries the date it was last checked.
My newsletter says In analyzer queue and never changes. What happened?
Most likely your weekly limit. An email over the limit keeps the amber chip and never gets an analysis, because there is no separate skipped state. Check Used this week in Settings; the counter resets on Monday. The other cause is a reading that could not be made sense of, which leaves the newsletter with an empty analysis.
Are the open rate and CTR figures real?
No. Newsletrix estimates both from the content of the email against generic benchmarks, and every analysis carries a note stating the basis of each estimate. Newsletrix cannot see the sender's email platform, so it cannot observe an open or a click. Use them to rank newsletters against each other, not as absolute numbers.
What is a workspace and do I need one?
A workspace is an optional label on a tracking address. Label one address Competitors and another Own sends, and the selector at the top of the sidebar scopes every chart to one of them. Without labels everything sits in All workspaces. Renaming or deleting a workspace changes only the label; it never deletes newsletters.
Which plan do I need, and what does upgrading change?
Every screen is available on every plan; plans differ in volume. Free gives 2 analyses a week and 1 tracking address, Basic 5 a week and 1 address, Standard 15 a week and 5 addresses, Premium 50 a week and 15 addresses, Founders 50 a week. Unlimited is an internal tier that cannot be purchased.