- Where
- Sidebar: Content / Structures
- App route
/products- Answers
- What blocks and products an email pushes
- Needs
- Analysed newsletters; 2+ sends for recurrence
- Updates
- Minutes after each analysis, with a short lag
What this screen is
The sidebar calls it Structures. The screen itself is titled Products & Categories, and both names fit, because Newsletrix reads two different things out of every analysed email and puts them side by side. One reading is structural. The other is about what the brand is selling.
The structural half answers what a brand puts in its emails, and in what order. Each send produces block names in reading order, such as hero, product_grid, category_cta, blog_teaser and footer. Join those with arrows and you have a flow. When the same flow comes back send after send, the app calls it a pattern. One pattern reused means a locked template. A new pattern every time means hand-built sends.
The merchandising half answers what they are selling, and how hard. Each product name is followed across everything you have collected, so a tile carries a total, the dates it ran between, and a warning when it comes back too often inside 30 days. All of it describes what a competitor chose to show. Newsletrix receives these emails as an ordinary subscriber, so nobody else's opens or clicks are visible here, and nothing on this screen tells you how the audience reacted.
Every value here is an AI reading of the email body. There is no date range picker, so you are always looking at your entire collected history.
Open it when
- You are rebuilding your own template and want the block order competitors actually ship.
- You want to know which product a brand is flogging hardest, and whether they are burning it out.
- You are planning a category push and want the categories your tracked set promotes most.
- You are choosing between product shots, lifestyle photography and banners, and want each brand's split.
- You need promoted products with price, discount and prominence for a merchandising review.
- A brand's emails feel busier and you want to check whether the block count really changed.
Screen anatomy
The real screen first, then a numbered map of the current layout. Every tab shares the same header and brand filter; only the panels below the tab bar change.
Products & Categories
Comprehensive analysis of promoted categories, section flow patterns, and image types across newsletters.
| Section flow pattern | Count |
|---|---|
| hero -> product_grid -> category_cta -> blog_teaser -> footer | 41 |
| hero -> product_grid -> footer | 27 |
| hero -> banner -> product_grid -> newsletter_signup -> footer | 14 |
| Brand | Section name | Section count | Newsletter subject |
|---|---|---|---|
| Be Lenka | hero -> product_grid -> footer | 3 | New season barefoot styles are here |
| Vivobarefoot | hero -> product_grid -> category_cta -> blog_teaser -> footer | 5 | Last chance: 30% off |
| Alfred | hero -> informational_text -> cta_block -> footer_and_social | 4 | The Gift That Always Fits |
- Brand filter A multi-select labelled "Brand", placeholder "All brands". It narrows the five tables, Top promoted categories, Image type usage by brand and both Promo cards. Four charts ignore it. Changing it resets every table to page 1.
- Tab bar Patterns (the default), Flow, Images, Promo, Categories. Switching tabs swaps everything below the bar and leaves the header and brand filter in place.
- Section frequency distribution The 20 most frequent block names across every analysed email. That is your market's vocabulary; one brand's template lives in the table further down.
- Most common section patterns The 10 most repeated block sequences, with how many newsletters used each one exactly.
- Section statistics One row per newsletter, newest first, carrying the brand badge, the whole arrow-joined flow, the block count and the subject.
- Export CSV One button per table. It downloads the rows shown on screen, so at most the 50 of the visible page.
- Section header and help icon Title, subtitle, and a help icon opening a modal called "Product Promotion Excellence".
- Pagination footer "Showing X to Y of Z results" with page buttons, only when there is more than one page. Page size is fixed at 50 and each table paginates independently.
- The other four tabs Flow lists one row per block with its position. Images holds a doughnut, a stacked bar per brand and an evidence table. Promo has the product tiles, the recurrence timeline, a prominence doughnut and the product table. Categories gives you the ranking, a Key insights card and a table.
- Loading and empty states A spinner sits at the bottom while a table fills. A chart with nothing behind it is left out altogether, so a tab simply looks shorter.
Every panel explained
Section frequency distribution (Patterns)
Horizontal bars of the most-used block names across your whole archive.
How it is calculated. Every block name is counted wherever it occurs, and the 20 highest are drawn, longest bar first. A block used twice inside one email counts twice.
How to read it. Long bars are the blocks in nearly every email, the short tail is experimentation. Names are free text, so hero and hero_banner sit on separate bars.
This chart ignores the Brand filter in the current release, and so do Most common section patterns, Image type distribution and Promo products by prominence. With a brand selected those four still describe your whole tracked set while the tables describe one sender.
Most common section patterns (Patterns)
The ten most repeated block sequences, with the number of newsletters behind each.
How it is calculated. A newsletter's pattern is its block names in reading order, joined with arrows. Identical lines count together and the ten biggest counts are listed. Matching is exact, character for character.
How to read it. A high count is a locked-down template; many counts of 1 means hand-built sends or inconsistent naming. One extra block creates a new pattern, which is the earliest sign of a redesign.
Section statistics (Patterns)
One row per newsletter, with the brand, the full block flow, the block count and the subject.
How it is calculated. The blocks of a single send are joined with arrows in reading order and counted. A newsletter with nothing found still appears, at a count of 0. Fixed to newest first.
How to read it. The fastest check of whether a send followed the house template. The header says "Section name", but the cell holds the whole flow.
Newsletter section flows drill down (Flow)
One row per single block, showing brand, subject, date, section order and the block name.
How it is calculated. Every extracted block listed on its own, newest send first, 50 rows a page.
How to read it. Use it when you need one block's exact position, for example where the blog teaser sits. Two lines that came out of one send sit together only because the table is ordered by date.
The first block in an email is 0, not 1. A block whose position is unknown also shows 0, so a genuine first block and an unplaced one look identical in that column.
Image type distribution (Images)
A doughnut of the visual styles detected, with the newsletters behind each slice in the tooltip.
How it is calculated. Each email contributes one entry per distinct image type it uses, and the slices count those entries by type. An email with twenty product shots contributes one product-shot entry.
How to read it. The dominant slice is your market's visual language. It ignores the Brand filter, so pair it with the stacked chart.
Image type usage by brand (Images)
Stacked bars, one per brand, one colour per image type.
How it is calculated. The same entries, grouped by brand and type, brands in alphabetical order, and a combination that never occurred sits at zero. A send whose brand could not be identified is left out.
How to read it. Height is visual volume, segments are style. This chart follows the Brand filter, so it can legitimately disagree with the doughnut.
Image types used in newsletters (Images)
Brand, subject, date and image type, one row per type per newsletter.
How it is calculated. A straight listing, newest first, 50 rows a page, blanks as "N/A".
How to read it. Use it to jump from a doughnut slice to the sends behind it.
Top products (Promo)
Up to eight tiles. Each one carries the product name, the appearance count, up to two brand badges, first and last seen, a discount chip, a price chip, a hero chip and an amber "Over-exposed" chip.
How it is calculated. Products are matched on name alone, ignoring capitals and stray spaces. Appearance count is distinct newsletters, first and last seen the earliest and latest send dates, hero count the high-prominence entries. Ranked by appearances then recency, top eight shown, and only the first discount and price wording appears on the tile.
How to read it. These are the products a brand has decided to bet on. "3 hero" means the product led three sends, not that it appeared three times.
A tile can only show "Over-exposed" if the product also reached the recurrence timeline, which keeps the twelve most frequent recurring products. Outside that list nothing is ever measured against the threshold, so a product with nine appearances can sit there unflagged.
Product recurrence timeline (Promo)
One row per recurring product, a dot per appearance on a shared date axis, with the total, the peak inside a 30-day window and an amber pill when the threshold is crossed.
How it is calculated. A product needs two or more appearances, and only the twelve most frequent are charted. The peak is the most appearances that fall inside any 30 consecutive days, and five or more raises the flag. Dot colour shows prominence, solid for high, faded for medium, grey for low or unknown.
How to read it. Clustered dots mean a burst campaign. An even spread is an evergreen hero, and an amber row is a repetition risk. The axis covers all history, so short-lived products look squeezed.
Promo products by prominence (Promo)
A doughnut of how many product entries were tagged high, medium or low prominence.
How it is calculated. Entries counted by prominence value, high first, then medium, low, and anything else; empty values become "Unknown".
How to read it. A large "high" share means a few hero products lead the emails, a fat "low" share means long grids with no clear lead. Prominence is whatever word describes the placement in that email, so an unexpected value gets its own slice. Ignores the Brand filter.
Promotional products in newsletters (Promo)
Brand, product name, description, discount, price, prominence, newsletter subject and date.
How it is calculated. Every product entry, newest first, 50 rows a page, blanks as "N/A". The prominence badge is red for high, yellow for medium, blue for everything else including empty.
How to read it. Check a tile before you quote it: which sends carried the product, and what offer sat next to it each time.
Price and discount are kept word for word as they appeared in the email, so "EUR 89.90", "-30%" and "2+1 free" share one column. No currency conversion, no arithmetic. For structured price analysis use the Promotions screen.
Top promoted categories (Categories)
Horizontal bars of the 15 most frequently promoted categories, labelled "Promotion Count".
How it is calculated. Every promoted category is counted, and the 15 highest are drawn, longest bar first. This chart follows the Brand filter.
How to read it. The category mix a brand pushes in email, as opposed to the mix on their website. Categories are free text, so "kids" and "kids shoes" occupy separate bars.
Key insights (Categories)
A card beside the category chart with up to five auto-written sentences, such as "kids" is the top promoted category with 21.4% of promotions.
How it is calculated. The card restates figures already on the screen around it. It reads off the top category and its share, the most frequent block name, the number of flow patterns, the dominant image type and its share, and the count of high-prominence products.
How to read it. Every share is a share of a shortened list, so a category percentage covers only the top 15 and the pattern count can never exceed 10. Three of the five panels behind the card ignore the Brand filter, so with one brand selected the card can describe your whole set while the table under it shows one sender.
Categories in newsletters (Categories)
A paginated table of brand, category, subject and date, one row per category per newsletter.
How it is calculated. A straight listing, newest first, 50 rows a page.
How to read it. Use it to date a category push. One newsletter promoting five categories produces five rows, so the result count runs above your newsletter count.
Metric definitions
| Metric | What it means | How it is produced | Scale |
|---|---|---|---|
| Frequency (section) | How often a block name occurs across your set | Every extracted block counted by name, top 20 kept | Whole number, 1 and up |
| Count (pattern) | How many newsletters used one exact block sequence | Block names joined in order into one line, identical lines counted, top 10 kept | Whole number, 1 and up |
| Section count | How many blocks were extracted from one newsletter | Count of that email's block entries; 0 when nothing was extracted | 0 and up |
| Section order | Position of a block inside its email | Reading order inside the email, counting from 0 | 0 and up |
| Section flow | The block sequence as one line of text | Block names joined with arrows in position order | Text |
| Promotion Count | How often a category was promoted | Every extracted category counted, top 15 kept | Whole number, 1 and up |
| Image type count | How many emails reported that visual style | One entry per distinct type per email, counted by type | Whole number, 1 and up |
| Newsletter count | Distinct emails behind a doughnut slice, tooltip only | Distinct newsletters counted for that type or prominence value | Whole number, 1 and up |
| Prominence | How prominently a product was featured | Read from the email; usually high, medium or low, though another word can appear | high / medium / low, another word, or empty |
| Appearance count | Distinct newsletters a product appeared in | Products matched on name alone, capitals and stray spaces ignored | Whole number, 1 and up |
| First seen / Last seen | When the product was first and last promoted | Earliest and latest send date among its appearances | Date |
| Peak N/30d | Most appearances inside any rolling 30-day window | The most appearances that fall inside any 30 consecutive days | 0 and up |
| Over-exposed | The product was pushed hard enough to risk fatigue | Peak reaching the fixed threshold of 5 inside 30 days | Yes / no |
| Hero count | Times the product carried high prominence | High-prominence entries counted for that product name | 0 and up |
| Price / Discount | The offer wording as it appeared in the email | Kept word for word, nothing converted and nothing tidied up | Free text or "N/A" |
Filters and controls
| Control | What it does | Default |
|---|---|---|
| Brand | Multi-select of the brands with at least one analysed newsletter, ordered by volume. Filters the five tables, Top promoted categories, Image type usage by brand and the Promo cards. Does not filter Section frequency distribution, Most common section patterns, Image type distribution or Promo products by prominence. Resets every table to page 1. | Empty, meaning all brands |
| Tabs | Switches between Patterns, Flow, Images, Promo and Categories. The brand filter and the page header stay put as you move between them. | Patterns |
| Pagination | 50 rows per page, independent per table. The footer and page buttons only appear when there is more than one page. | Page 1 |
| Export CSV | One button per table. Downloads the rows visible on screen plus the header row. | Current page only |
| Help icon | Opens the "Product Promotion Excellence" modal describing what the tab covers. | Closed |
| Sorting | Fixed. Every table is ordered by send date, newest first, and there is no sort control to change it. | Newest first |
Four things people go hunting for are simply not here. No date range picker, no text search, no column sorting. The group and tag filter used elsewhere in the app does not reach this view either. Whatever you do, the screen covers your entire collected history.
How to use it
- Start on Patterns with no brand selected. Read the frequency chart as vocabulary: which blocks exist in your market and which are near-universal. A name appearing once is usually the AI wording a familiar block differently.
- Pick one brand and read Section statistics, not the charts. The frequency chart and the pattern table stay market-wide. The table is where a single sender's flows and block counts live, one row per send.
- Lift the template. Take the highest-count pattern as your block skeleton, then switch to Flow for the exact position of a specific block, remembering that positions count from 0.
- Check the visual language on Images. The stacked chart is the filtered per-brand comparison, the doughnut is market context. If they disagree with a brand selected, that is the filter difference, not a bug.
- Work the Promo tab for merchandising. Top products shows the bets. The recurrence timeline shows the burn risk, five or more appearances inside 30 days, and the product table carries the send-by-send evidence.
- Finish on Categories and export. Since an export only carries the 50 rows on screen, narrow to one brand first so the visible page is the page you want.
Screenshot a brand's top pattern once a quarter. Pattern matching is exact, so a template redesign shows up as a brand new sequence with a count of 1 while the old pattern stops growing. That contrast is the earliest structural signal you get, usually weeks before the change is obvious by eye.
Limits and caveats
- Everything here is an AI reading of the email body, not a measured metric. A missed block or an invented block name travels straight into the charts.
- In the current release the Brand filter does not reach Section frequency distribution, Most common section patterns, Image type distribution or Promo products by prominence.
- Every ranking stops at a fixed length. 20 block names, 10 flow patterns, 15 categories, 8 top products, 12 recurrence rows. Nothing is swept into an "other" bucket, so a Key insights share is a share of the shortened list.
- Products are matched on name alone, so a rename, a size suffix or a typo splits one product into several.
- The recurrence timeline has no time window and spans all history, so a long axis can make a recent burst look narrow.
- Figures here can sit a few minutes behind a fresh import, so a newsletter analysed moments ago may not have reached every panel yet. The brand list is usually the last to catch up.
FAQ
Why did the section frequency chart not change when I picked a brand?
In the current release, four widgets always show every brand you track, whatever the filter says. They are Section frequency distribution, Most common section patterns, Image type distribution and Promo products by prominence. The Brand filter does reach the five tables, Top promoted categories, Image type usage by brand and both Promo cards. Read those four unfiltered widgets as market-wide context, and use the tables when you need one brand.
What exactly is a section?
A layout block name the AI produced for that specific email, listed in reading order. Typical names are hero, product_grid, category_cta, blog_teaser and footer. There is no fixed vocabulary, so two brands can use different words for the same block, and one brand can drift over time.
Why does one newsletter appear on many rows?
The Flow, Images and Categories tables list one row per extracted item, so every block, every image type and every category gets its own line. Only Section statistics collapses to one row per newsletter. That is why the result counts under those tables run much larger than your newsletter count.
What makes a product Over-exposed?
The same product name, ignoring capitals and stray spaces, appeared in five or more newsletters inside any rolling 30-day window. Five appearances and 30 days are the fixed defaults, and the card prints the exact peak next to the flag as peak N/30d.
Why does a product with nine appearances carry no over-exposure badge?
The 30-day peak is only tracked for products that also reach the recurrence timeline, which is limited to the twelve most frequent recurring products. Outside that list nothing is measured against the threshold and the tile can never be flagged, however high its total appearance count is.
Can I use the Price and Discount columns for a pricing analysis?
Treat them as quotes rather than numbers. They are the wording that appeared in the email, with no currency conversion and no tidying up, and the Top products tile shows only the first of several discount wordings. For structured price work use the Promotions screen.