Skip to content

What's inside the newsletter from Connox

Sender
shop@connox.fr
Sector
Home & decor
Emails read
55 campaigns, 6 transactional
Period
August 2026
Overall score

The main finding

The DKIM signature was not validated on all 6 transactional messages.

It is the first of 6 points to fix. The highest score is compliance (100/100), the lowest deliverability (54/100), better than 4% of the 579 brands measured.

Monitor Connox See the 6 points

One email when a criterion changes state: the Pro version, in early access.

PDF
Share

Share this report

A single message, written by us. Their address is used for this message only and is not added to any list.

6 points to fix

Each line leads to the criterion in its family, with the excerpt that establishes it and how to fix it.

  1. Priority The DKIM signature was not validated Criterion “DKIM signature validated” Deliverability on all 6 transactional messages −20 pts out of 20

    DKIM signing is set up in the email platform (“domain authentication”): it provides a CNAME or TXT DNS record to publish on the sending domain, then checks the publication.

  2. To review Preheader missing, too short or too long Criterion “The message makes sense without its images, from the inbox on” Experience measured on 6 transactional messages −12.5 pts out of 25

    A preheader of 40 to 130 characters, in real text, follows the subject in the inbox; every meaningful image has an alt attribute that says what it shows; the main message is in text, not inside an image.

  3. To review The message puts obstacles in the way of assistive technology Criterion “The message stays readable with assistive technology” Experience measured on 6 transactional messages −10 pts out of 20

    Four fixes, all in the template: declare the message language on the <html> tag (lang="en" for an email in English), describe each content image in its alt attribute, mark layout tables with role="presentation", and write links that say where they lead rather than “click here”.

  4. To review Some links lead nowhere Criterion “Every link leads somewhere” Technical on all 6 transactional messages −10 pts out of 10
  5. To review The published DMARC policy keeps out no impersonation Criterion “DMARC policy that helps keep out some impersonation” Deliverability on 4 of the 6 transactional messages −10 pts out of 15
  6. Secondary No doctype declared Criterion “Well-formed HTML” Technical measured on 6 transactional messages −3.8 pts out of 15

15 criteria met, including « No open-tracking pixel » and « DMARC alignment validated ». They are listed in their family, below.

Every send, at the time it arrived

One dot per message received by our subscription since August. Tap a dot to open it; the arrows move from one send to the next.

Technical reading of the send of Aug. 21, 2026 →

campaigns in 11 weeks
55
about 5 a week, from August to October
every day of the week
7 days a week
4 to 11 campaigns depending on the day
favourite sending hour
10 h
85% of campaigns arrive between 10:00 and 11:00
last send received
31 Oct.
« Neuheiten im Oktober 🍁 »
Pixel
Score
55 48 52 53 53

Campaign 55 of 55

Saturday 31 Oct., 10:14

Neuheiten im Oktober 🍁

Type
Marketing campaign
Score
analysis in progress
Pixel
—
View the preview Technical reading →

Sending day

  1. Mon.
  2. Tue.
  3. Wed.
  4. Thu.
  5. Fri.
  6. Sat.
  7. Sun.

Arrival time

3 changes observed since August

Each campaign is compared with the previous one, criterion by criterion.

15 Aug.

The detail of this change, criterion by criterion, is reserved for the Pro version.

confirmed by the next campaign · the per-send score goes from 55 to 48

12 Oct.

The detail of this change, criterion by criterion, is reserved for the Pro version.

confirmed by the next 5 campaigns · the per-send score goes from 48 to 52

20 Oct.

The detail of this change, criterion by criterion, is reserved for the Pro version.

confirmed by the next 8 campaigns · the per-send score goes from 52 to 53

See what changed, before and after Version Pro

Criteria in detail

The 24 criteria + 1 bonus

Met The criterion is verified on every email where it applied. Counts for its full weight in the score. Partly met The criterion is verified on only some of the emails; the detail says which. Counts in proportion to the emails where it is met. Not met The criterion is verified on none of the emails where it applied. Counts as zero, and its weight stays in the denominator. Not observable Nothing in the emails received allows a conclusion: the criterion leaves the denominator and is never counted as a failure. Leaves the denominator: listed, never counted as a failure. Not applicable The criterion does not concern these emails: the question does not arise there — an email type that does not call for it, a clickable area with no declared size. Distinct from “not observable”, which says the measurement could not be made. Leaves the denominator: the question does not arise for this type of email.

Compliance

These checks cover what the send imposes on the recipient and what it leaves them: measurement of their behaviour, ways to leave, tracking of their objection.

Tracking pixels: the CNIL rule Method, Compliance family better than 98% of the 580 brands measured

The calculation · 100 pts out of 100

5 met · 1 not applicable

35 20 20 15 10

Weight of each criterion in the score.

No open-tracking pixel Met No open-tracking pixel on the last 3 transactional messages 35 pts

35 points out of 35 — met on 3 of the last 3 transactional messages.

Met when no tracking pixel is present. A single pixel is enough for the criterion not to be met on an email.

Unsubscribing is possible and announced Not applicable not applicable

This measurement did not apply to any of the emails received. It drops out of the denominator: it is never counted as a failure.

Met when the message announces unsubscribing in its headers, through a page rather than an email to write, and in its body. The detail says which of the two is missing. Not applicable to a transactional email.

  • RFC 2369
  • PM-PRIV-002
The recipient keeps control over tracking Met The recipient keeps control over tracking on 2 transactional messages 20 pts

20 points out of 20 — share of observations met across 2 transactional messages.

Two measures: does the unsubscribe link go through a tracker, and does a separate path lead to a preference center or to refusing tracking? Neither establishes that a refusal succeeds — checking would require following the link, which would record a click on the sending platform's side. They describe the control left to the recipient, not the result they would get.

What was measured — 1 met out of 1

  • Untracked unsubscribe link on 2 transactional messages Met
  • A separate path leads to a preference center or to refusing tracking Not applicable
  • CNIL 14/04/2026 · Postmetry
  • PM-PRIV-003
The message says who is writing and under which rules Met The message says who is writing and under which rules on 6 transactional messages 20 pts

20 points out of 20 — share of observations met across 6 transactional messages.

A commercial email must make it possible to identify the advertiser (CPCE, article L. 34-5), and the recipient must be able to know who processes their data. This indicator groups what a message can show of it.

What was measured — 3 met out of 3

  • The message names its sender on 6 transactional messages Met
  • A reply to the message is possible on 6 transactional messages Met
  • A link leads to the privacy policy on 6 transactional messages Met
  • CPCE L. 34-5 · directive 2000/31 art. 6 · CAN-SPAM
  • PM-PRIV-004
An untracked path leads to the site Met An untracked path leads to the site on 6 transactional messages 15 pts

15 points out of 15 — met on 6 of 6 transactional messages concerned.

Met when at least one link in the message leads to the brand's site without going through a click counter.

  • Postmetry
  • PM-PRIV-005
No recipient identifier in the message Met No recipient identifier in the message on 6 transactional messages 10 pts

10 points out of 10 — met on 6 of 6 transactional messages concerned.

Met when the recipient's address appears — neither in clear, nor encoded, nor hashed — in no URL and no proprietary header. Addressing headers do not count: they carry it by design.

  • Postmetry · RGPD art. 4 en repère
  • PM-PRIV-020

Deliverability

These checks read the authentication verdict set by our server when the message arrived. This verdict is dated by construction: we never recalculate it, which would query today's DNS rather than the DNS of the day the email was sent.

Gmail sender requirements Yahoo requirements Method, Deliverability family better than 4% of the 579 brands measured

The calculation · 35 pts out of 65, scaled to 100

2 met · 1 partly met · 1 not met · 2 not applicable

20 20 15 10

Weight of each criterion in the score.

DKIM signature validated Not met DKIM signature failed on the 6 transactional messages concerned 0/20 pts

0 points out of 20 — met on 0 of 6 transactional messages concerned.

How to fix it

DKIM signing is set up in the email platform (“domain authentication”): it provides a CNAME or TXT DNS record to publish on the sending domain, then checks the publication. A failure on a signed message most often comes from content changed after signing (a footer added by a gateway, rewritten links) or from a key removed from DNS.

Without a valid signature, Gmail and Yahoo have treated bulk mail as unauthenticated since 2024, and reject it or file it as spam.

Met when our receiving server validated the DKIM signature of the email.

DMARC alignment validated Met DMARC alignment verified on 6 transactional messages 20 pts

20 points out of 20 — met on 6 of 6 transactional messages concerned.

Met when the domain shown in the sender matches the one that authenticated the message.

DMARC policy that helps keep out some impersonation Partly met DMARC policy in observation only (4 of 6 transactional messages concerned) 5/15 pts

5 points out of 15 — met on 2 of 6 transactional messages concerned.

How to fix it

A _dmarc.<domain> TXT record declares the policy: v=DMARC1; p=quarantine; rua=mailto:… at the very least, p=reject once the sending sources are aligned. The move happens in stages (p=none with reports, then quarantine, then reject) over a few weeks, reading the reports to find the tools that send on behalf of the domain.

A none policy prevents no spoofing of the domain; quarantine or reject closes the door and meets the Gmail and Yahoo requirements for bulk senders.

Met when the published policy is “reject” or “quarantine”. A “none” policy asks nothing of mail providers: it is published with no effect on messages that impersonate the domain.

No SPF failure Met No SPF failure on 6 transactional messages 10 pts

10 points out of 10 — met on 6 of 6 transactional messages concerned.

Met when SPF does not fail. An SPF “none” alongside a valid DKIM and DMARC is a common and legitimate configuration: it is not penalized.

One-click unsubscribe, as defined by RFC 8058 Not applicable not applicable

This measurement did not apply to any of the emails received. It drops out of the denominator: it is never counted as a failure.

Met when the “List-Unsubscribe” header carries an https address and “List-Unsubscribe-Post” is exactly “List-Unsubscribe=One-Click”. This is what Gmail and Yahoo require of bulk senders: the mailbox's button then acts without opening a page. Not applicable to a transactional email.

  • RFC 8058 · Gmail et Yahoo
  • PM-DELIV-005
A DKIM signature that covers what matters Not applicable not applicable

This measurement did not apply to any of the emails received. It drops out of the denominator: it is never counted as a failure.

What the signature covers, not only that it passes: an algorithm accepted by RFC 8301, the whole body signed, the From, Subject, Date and List-Unsubscribe headers signed, a signature not expired on receipt. Not applicable without a DKIM signature.

What was measured — 0 met out of 0

  • Accepted signing algorithm Not applicable
  • Whole message body signed Not applicable
  • Essential headers signed Not applicable
  • Signature valid on receipt Not applicable
  • RFC 6376 · RFC 8301
  • PM-DELIV-006

Technical

These checks cover what email clients actually do with the message: truncation, style blocks removed, resources never loaded. Each threshold states where it comes from: the documented behaviour of an email client, a specification, or our own method.

Support by email client (caniemail) Method, Technical family better than 49% of the 580 brands measured

The calculation · 71 pts out of 85, scaled to 100

4 met · 1 partly met · 1 not met

25 20 15 10 10 5

Weight of each criterion in the score.

HTML under Gmail's clipping threshold Met HTML under Gmail's clipping threshold on 6 transactional messages 25 pts

25 points out of 25 — met on 6 of 6 transactional messages concerned.

Met when the HTML part stays under 102 KB. Beyond that, Gmail cuts off the end of the message — and the end is where the unsubscribe link and the notices are.

  • Gmail
  • PM-TECH-001
A shallow template Met Shallow template on 6 transactional messages 20 pts

20 points out of 20 — share of observations met across 6 transactional messages.

Met when layout tables are nested no more than 6 levels deep. Beyond that, the Word engine shared by desktop Outlook and the diverging clients each render the template their own way.

What was measured — 1 met out of 1

  • Lightly nested layout tables on 6 transactional messages Met
  • Postmetry
  • PM-TECH-002
Well-formed HTML Partly met No doctype declared on the 6 transactional messages concerned (3 of 4 measures met) 11.2/15 pts

11.2 points out of 15 — share of observations met across 6 transactional messages.

How to fix it

Form errors (unclosed tags, duplicated attributes, invalid table/div nesting) come from blocks pasted between editors or from hand edits in the code. Run the HTML through a validator before sending, and fix it at the source in the template rather than in each campaign.

Malformed HTML renders differently from one email client to another, and it is the most common cause of a broken display in Outlook or on mobile.

A doctype, a character set declared in the document, balanced tags and unique identifiers. Each client repairs incomplete markup its own way, which is what makes rendering unpredictable. End tags that HTML makes optional do not count.

What was measured — 3 met out of 4

  • No doctype declared on the 6 transactional messages concerned Not met
  • Character set declared in the HTML on 6 transactional messages Met
  • Balanced tags on 6 transactional messages Met
  • Unique identifiers on 6 transactional messages Met
  • WHATWG
  • PM-TECH-030
Every link leads somewhere Not met Some links lead nowhere on the 6 transactional messages concerned 0/10 pts

0 points out of 10 — met on 0 of 6 transactional messages concerned.

How to fix it

Every link has an absolute https target; no href="#", no href="", no relative link and no mailto: without an address. Empty links most often come from a block copied from an earlier template, or from a button whose URL was never filled in in the editor.

An empty link loses the click and signals a message put together in a hurry; the fix is a review of the link list before sending, which most editors offer.

Met when no link is empty, reduced to “#” or written as “javascript:”. Such a link leads nowhere, and most clients neutralise it. Not applicable if the message contains no link.

  • Postmetry
  • PM-TECH-031
No remote stylesheet Met No remote stylesheet on 6 transactional messages 10 pts

10 points out of 10 — met on 6 of 6 transactional messages concerned.

Met when the message calls no external stylesheet, through “link” or “@import”. Several clients never load them. A font stylesheet does not count: without it, the text falls back to another font.

The message follows the shape of an email Met The message follows the shape of an email on 6 transactional messages 5 pts

5 points out of 5 — share of observations met across 6 transactional messages.

Lines of 998 bytes at most, readable Message-ID and Date, a character set announced on every text part, a transfer encoding consistent with the content (RFC 5322, RFC 2045). Most servers tolerate these deviations, each in its own way.

What was measured — 5 met out of 5

  • Lines of 998 bytes at most on 6 transactional messages Met
  • Well-formed Message-ID on 6 transactional messages Met
  • Readable date on 6 transactional messages Met
  • Character set announced on every text part on 6 transactional messages Met
  • Consistent transfer encoding on 6 transactional messages Met
  • RFC 5322 · RFC 2045
  • PM-TECH-010

Experience

Four questions about what becomes of the message once received: does it stay understandable, can the recipient still act, can a client render it, can a screen display it. Automatic detection only covers what is declared in the message.

WCAG 2.2 Method, Experience family better than 88% of the 580 brands measured

The calculation · 78 pts out of 100

4 met · 2 partly met · 1 bonus

25 15 15 10 20 15

Weight of each criterion in the score.

The message makes sense without its images, from the inbox on Partly met Preheader missing, too short or too long on the 6 transactional messages concerned (1 of 2 measures met) 12.5/25 pts

12.5 points out of 25 — share of observations met across 6 transactional messages.

How to fix it

A preheader of 40 to 130 characters, in real text, follows the subject in the inbox; every meaningful image has an alt attribute that says what it shows; the main message is in text, not inside an image. In the editor, the preheader is a dedicated field, and the alt is filled in image by image.

Half of readers see the message without images when it first opens. An alt and a preheader take a minute per campaign.

Met when the message keeps at least 200 readable characters with images blocked, and a preheader of 40 to 130 characters announces it in the message list. Without a preheader, the inbox shows the first text of the body instead — often “View in browser”.

What was measured — 1 met out of 2

  • At least 200 characters remain readable with images blocked on 6 transactional messages Met
  • Preheader missing, too short or too long on the 6 transactional messages concerned Not met
  • Postmetry
  • PM-EXP-001
The message can be rendered outside HTML Met A text version renders the message on 6 transactional messages 15 pts

15 points out of 15 — share of observations met across 6 transactional messages.

Met when a text version comes with the email and carries the essentials: at least 60% of its text and 80% of its links. A client that does not display HTML falls back on it; reduced to “view the web version”, it renders nothing.

What was measured — 3 met out of 3

  • A plain-text version accompanies the message on 6 transactional messages Met
  • The text version carries at least 60% of the text on 6 transactional messages Met
  • The text version carries at least 80% of the links on 6 transactional messages Met
  • Postmetry
  • PM-EXP-002
All links stay visible with images blocked Met All links stay visible with images blocked on 6 transactional messages 15 pts

15 points out of 15 — met on 6 of 6 transactional messages concerned.

Met when no link is carried by a single image without alternative text: such a link leaves nothing on screen. Not applicable if the message contains no links.

  • Postmetry
  • PM-EXP-003
Tap targets large enough Met Tap targets of sufficient size on 6 transactional messages 10 pts

10 points out of 10 — met on 6 of 6 transactional messages concerned.

Met when no sized tap target falls below 24 px a side (WCAG 2.2, criterion 2.5.8). Only targets with a declared dimension are measured; the criterion's exceptions — a target within a line of text, sufficient spacing — are not. Not applicable if the message declares no dimensions.

The message stays readable with assistive technology Partly met The message puts obstacles in the way of assistive technology — 2 of 4 measures met 10/20 pts

10 points out of 20 — share of observations met across 6 transactional messages.

How to fix it

Four fixes, all in the template: declare the message language on the <html> tag (lang="en" for an email in English), describe each content image in its alt attribute, mark layout tables with role="presentation", and write links that say where they lead rather than “click here”.

The criterion detail shows which ones are missing from your emails; the others are already met.

A decorative image — divider, spacer — keeps an empty alt="": that is what keeps it silent, not an oversight.

Declared language, described content images, layout tables marked as such, link texts that say where they lead: what a screen reader needs to render the message. The detail says which of these measures is missing.

What was measured — 2 met out of 4

  • Message language not declared on the 6 transactional messages concerned Not met
  • Every content image is described on 6 transactional messages Met
  • Layout tables announced as data tables on the 6 transactional messages concerned Not met
  • Every link says where it leads on 6 transactional messages Met
The message declares its adaptation to screens Met The message declares its adaptation to screens on 6 transactional messages 15 pts

15 points out of 15 — share of observations met across 6 transactional messages.

Met when the message declares its viewport and its adaptation to screen width. Without them, it displays on a phone at its original width, and is read by zooming and scrolling sideways.

What was measured — 2 met out of 2

  • Viewport declared on 6 transactional messages Met
  • Adaptation to screen width declared on 6 transactional messages Met
Dark mode declared Bonus point not earned The message does not declare its support for dark mode on the 6 transactional messages concerned +0 pts

0 points out of 5 — met on 0 of 6 transactional messages concerned.

Earned when the message declares color-scheme or a prefers-color-scheme rule. No standard requires it — clients that force dark mode invert the colors themselves — but declaring it leaves control to the template rather than to the mail client.

  • caniemail
  • PM-EXP-027
Does this address really send Connox's emails? SPF pass · DKIM not evaluated · DMARC pass · 6 transactional

Across 6 transactional messages from Connox received at our addresses since August 2026:

From field
shop@connox.fr · 4 of 6 shop@connox.de · 2 of 6
Verdict on receipt
SPF pass · 6 of 6 DKIM not evaluated · 6 of 6 DMARC pass · 6 of 6

The verdict is the one the receiving server wrote in the Authentication-Results header when the message arrived: it is read, never recomputed.

These findings describe the emails received at our addresses. They say nothing about a message received elsewhere: the From field can be written freely, and only the headers of the message you received show the path it took.

Check a suspicious email →

Sector comparison

Home & decor — 88 brands with at least one campaign analyzed. The shares below are counted on them, never on the catalog.

88 brands analyzed in this sector Reserved for the Pro version

Reading notes

How to read this page

  1. A criterion that is not observable or not applicable leaves the calculation. It stays listed, never counted as zero: its weight leaves the denominator rather than weighing as a failure.

  2. Each line describes an observable finding, with the excerpt that establishes it. No legal qualification is ever attached to it.

  3. The measurements come from deterministic analyzers: the same message always gives the same result, and a disputed finding can be redone identically from the archive.

  4. Every share is published on an observed basis, and names it: “26 of 30 brands analyzed.” A brand with no email received enters no denominator.

  5. The deliverability score is borrowed: its verdict comes from the server that received the message. All the other families come from our own analyzers.

Does this analysis speak to you?

Let's go through it together, in 20 minutes.

We go over the 6 points with you, the evidence for each and the order in which to fix them. By video call, no commitment.

Pick a time slot Cal.com calendar

Or take it with you: the PDF report

Is this analysis useful to you?

One click, and you help us make it better.