What shipped

Very long pages are now captured, scored and checked

A page whose full-page screenshot came out taller than 16,384 pixels, most often a long page in its mobile view, failed to save every day, and Page Scores and Content Health skipped that page on that device along with it. Those screenshots are now saved and compared, so the page starts showing visual history, Lighthouse scores and link and image results for that device from its next scan.

• Full-page screenshots are now stored with no compression loss. Compression noise no longer counts towards a change, so the percentage on a small change can differ a little from earlier days, and a very faint colour shift that only registered through that noise may no longer be reported.

The counts beside each monitor now count sites

In an account with two or more sites, the red and amber counts beside Visual Changes, Uptime, Page Scores, Renewals and Content Health now say how many sites need a look, where some of them counted every page, link or image. Expect smaller numbers: a Content Health count of 56 and 27 can become 3 and 2 with nothing on the sites having changed.

• Beside each page title the counts are spelled out, such as "2 sites with broken pages" or "1 site with expected changes".
• Hover a count for the rest: the sites that couldn't be verified or are not checked yet, and the size behind the sites, such as how many broken pages there are in all or the soonest renewal deadline.
• A site whose latest Content Health scan failed, or one with expected visual changes and a page that couldn't be verified, is no longer counted as red or amber. The hover lists it as couldn't be verified.
• An account with one site sees no change.

Reports no longer show "Couldn't verify" under Renewals

When a renewal check has no reading to show, for instance because a registry such as .de or .at does not publish domain expiry dates, your client's report showed a dash over "Couldn't verify". That card is now left out of the report email and the shared report link, and the Renewals page in your dashboard still says which reading is missing and why.

• The card that has a reading spans the section under its own name, "SSL Certificate" or "Domain Expiry", instead of sitting beside an empty half, and the report preview draws it the same way. When neither card has a reading, the report has no Renewals section.
• Reports you shared before today follow the same rule when their link is opened.
• The registrar line is the registrar's name alone. A web address that a registry writes into that field is taken out, and a domain with no registrar on record has no registrar line where it used to read "Unknown". The Renewals page in the dashboard follows from each site's next daily check.
• The uptime calendar on the shared report link drops the same wording: hovering a day when the site turned every check away reads "no data, the site turned the checks away".

Cookie banners and chat widgets are hidden from screenshots again

Since June 16 the step that hides cookie banners, chat launchers and popups before a page is captured was not taking effect, so they sat in screenshots as part of the page. They are hidden again, now only where the capture recognises the tool behind them: consent tools, chat widgets, popup builders, Google ad units and accessibility launchers.

• Anything built into the site itself stays in the screenshot: a hand-built cookie bar or popup, a theme's announcement bar, a floating WhatsApp or quote button. The capture cannot tell these from real content, and hiding real content is the worse mistake.
• When a popup locked the page from scrolling, the sections further down could be captured empty. The page is now unlocked before it is captured.
• The next scan of a site with one of these may report a visual change once, because the banner is in the previous screenshot and not in the new one. Whether that sends an alert depends on the alert rules you have set.

Content Health alerts no longer repeat problems you were already told about

After a check that did not get through a page, because it was cut short, blocked or rate-limited, or because only its mobile check ran, the next alert could announce that page's long-broken links and images as new. Each page is now compared with the last check that actually read it, however long ago that was.

• A link that broke on a page whose previous check had not run was never announced: that day there was nothing to compare it with, and the day after it counted as old. It is now announced, once.
• When an alert still has nothing to compare with, it leaves out what you were already told about, says how many it left out, and adds the total to its subject line.

Content Health in reports is now a short checklist

In the report email, on the shared report link and in the report preview, Content Health is now three rows, Links, Images and Secure connections, each with a status icon, how many were checked and a short result. Your client no longer sees web addresses, technical failure reasons or a list of items that couldn't be verified.

• Each row compares with the last period in words, such as "down from 3", "up from 1" or "none last period", instead of an arrow beside one total.
• When a period has no usable scan, the section is left out of the report instead of showing a grey "Not checked this period" card.
• On a phone, the report email stacks the section into one column.
• Reports you shared before today also show the checklist, without their lists of addresses and without a checked count.

Reports show links and files that skip a secure connection

A new Secure connections row in Content Health counts links and images that still use plain http, in yellow. Scripts and stylesheets on plain http, which browsers block, turn it red.

The links and images checked count will drop

A link that appears on every monitored page, such as one in the menu, used to be counted once for each page. It is now counted once per site, so the figure falls noticeably. The dashboard shows it from each site's next scan, and reports from the first period scanned after today.

Report notices say when a recipient's address bounced

When a report could not reach some recipients, the notice to you said to resend their invitation, even for an address that had bounced or marked the report as spam, where a new invitation does not help. It now lists those addresses separately and says why. A confirmed address that bounces is also counted in the "didn't reach anyone" notice, which has its own subject in that case.

Alerts name each measure and device in words

Page Scores alerts in email, Slack, Teams and your alert history now read "Largest Contentful Paint (Desktop): 3.4 s" where they read "lcp: 3400". Visual Changes alerts say "/pricing (Desktop)" instead of a screen width, and a Page Scores run in the dashboard uses the same full names on its tiles.

The sidebar shows how many setup steps are left

Beside Portfolio, an amber number counts the setup steps still open, such as alerts or branding, and hovering over it names them. It goes away once every step is done or you dismiss the welcome card, and the card now drops a step you finish in settings without a reload.

A challenge page from Amazon's firewall is no longer mistaken for the site

When Amazon's firewall, AWS WAF, answers a check with a challenge page instead of the site, the status code that comes with it says nothing about the site, yet Content Health and Uptime read it as the site's own answer. Both now read it as "Couldn't verify".

Links to JavaScript-only pages move from Broken to Couldn't verify

A JavaScript-only page can answer with an error code even though it loads in a browser, and a link to one was counted as broken in Content Health, in your client's report and in alert emails. From each site's next daily scan it reads "Couldn't verify", labelled "Needs JavaScript".

Uptime asks a third location before a site that blocks one is marked down

A site that turns away the uptime check from one location could still be marked down, with an alert and an incident, on a single failure from a second location. A third location is now asked first.

The Content Health comparison in reports no longer runs into one word

The change beside the count of problems, such as "↓ -2 vs last period", was squeezed into one word, "↓-2vslastperiod", in the report email and on the shared report link. It now sits on its own line under the count.

Each monitor now has one page listing every site

Visual Changes, Page Scores, Content Health, Uptime, Renewals and Pages each list every site you monitor, worst first, and a row opens that site's detail in a side panel, or its screenshot history for Visual Changes. The site picker is gone, and your days of monitoring left sit where it was.

• In the sidebar, each monitor and Pages show their counts, red for problems and amber for warnings.
• Sites are added from Portfolio, and a site's own page opens from its row there or from an alert email.
• A domain you monitor for two clients opens that client's copy from Portfolio and from alert emails, and asks which client when a link does not say. It could open the other client's copy before.

Page Scores shows mobile and desktop separately, not averaged

The Page Scores list and each site's detail show one device at a time, mobile first, with a switch to desktop, where the page used to lead with scores averaged across both. A site that scores lower on mobile reads lower here than it did.

• The report your client receives is unchanged and still scores both devices together, so its figures can differ from those on Page Scores.
• A page with no score in the last seven days is listed as "Not scored yet". It used to be counted as "Healthy".

Removing a site no longer closes your client's other report links

Since September 9, removing one site from a client also closed the shared report links of every other site under that client, so a link you had sent could stop opening. Only the removed site's links close now, though links already closed this way stay closed.

Reports and the dashboard use one set of health colours

Red means a problem, amber a warning or a score of 50 to 89, and green a good result, the same in the report email, its PDF, the shared report link and the dashboard. Scores of 50 to 89 used to be orange in reports.

• Coloured words and numbers in the report use a darker shade of their colour, so they read clearly on white, and amber days on the uptime calendar carry dark dates instead of white ones.
• On the shared report link, the "Content update" label is the same blue as in the email. It used to take your agency colour, which a pale colour made hard to read.

Visual Changes gives fewer false readings

Two faults could make a comparison report the wrong thing: a capture that missed part of the page, and specks too small to be a change.

• A capture taken before the page's images had loaded, or while far more of it was moving than the day before, could report almost no change and then become the reference for the next day's comparison. It now reads "Couldn't verify", sends no alert and is never used as the reference.
• Specks a few pixels wide, left when text is drawn a fraction of a pixel differently, were boxed as edits and could lead the list of changes. They are dropped now unless the text itself changed.

A site no longer ends up checking fewer pages than it should

Two faults could leave a site with pages missing from monitoring.

• A page that answered "too many requests" while the site's pages were being looked up was taken off the list as though it had gone. A rate limit means "not now", so the page stays.
• When finding a new site's pages failed outright, the site was left with no pages to check, though it was still charged. The homepage is always on the list now, so its scans check at least that.

Sites keep running when the nightly charge falls behind

If the nightly charge fell behind, sites could stop, and an account with credits left could be told it was out of credits. A site now stops for credits only when your balance cannot pay for it.

• A paused site shows as paused from the moment you pause it. It used to read as still measuring until the next day.
• When a site comes back after a pause, or after running out and a top-up, the site's page names the dates nothing was watched, so you can answer a client who asks about the gap. That notice was meant to appear before and never did.

Portfolio opens a monitor's detail without leaving the list

The Mobile, Desktop, Assets, Uptime, SSL and Domain cells on Portfolio open that site's detail in place, and Back returns you to the same spot in the list. The Assets cell also shows problems and insecure items side by side, such as 4 + 23, so it adds up to the count on Content Health.

The dashboard says when the next scan and the next report are due

Outside a site's own page, the strip across the top of the dashboard now says when the next scan starts and when your next scheduled report goes out, where it used to say only "No scans running" between scans.

The Team page covers your whole agency

Every open invitation shows on the Team page and can be cancelled there. One sent while another client's site was selected did not show, and cancelling an invitation could say it was done while it stayed open.

• An invitation always adds a manager. To make someone an owner, change their role on the Team page once they have joined.
• The invitation email names your agency. It used to name whichever client was selected, which is not what the person joins.
• Someone who accepts an invitation goes straight into the dashboard instead of a "Select a site" screen.

The Manage recipients button in delivery warnings goes somewhere

When a report reached nobody, or skipped some of its addresses, the email telling you so ended in a Manage recipients button that opened a missing page. It now opens your settings at that client, where its recipients are edited.

A site you add now gets its first scan within minutes

Adding a site, at signup or from the dashboard, books its first scan for the next free quarter-hour, and the dashboard shows the time it will run. A site added from the dashboard used to wait for its own daily time, and for the following day when that time had already passed.

• The certificate and domain expiry are read on that first scan too, instead of up to a day later.
• A long list of sites is spread over the quarter-hours that follow, each row showing its own time.

Each site in the portfolio shows what that site costs

Every site under a client printed that client's whole monthly cost beside its own name, so a client with eight sites read as eight times the real figure. A row now carries what that one site costs, and the client's total moved onto the client name as hover text, which says how many sites it covers.

• Under the Portfolio title, the page now states how many sites you have, how many monitors are switched on, and what they come to a month.
• Add sites no longer totals the batch in its footer. It still warns when your balance will not cover everything you ticked, and the portfolio states the figure once the sites are in.

Reports can be sent from your own domain

Publish five DNS records for a domain you own and your reports leave from an address on that domain, signed by it, with no "via siteletter.com" line under your agency name. Setup sits in Settings under Report preferences: each record is listed with its own status, there is a Check now button, and you can cancel the setup, or stop sending from the domain later, at any time.

• On a verified domain the report body drops the SiteLetter logo and wordmark, whether or not you have set your own branding.
• A client's reply reaches your sending address when you have not filled in a reply-to of your own, instead of the address the account was opened with.
• This is not full white-label. Alert emails still come from SiteLetter, and a shared report link is still a siteletter.com address.

A report can go to your client, not only to your own inbox

The addresses you could put on a report were the people on your own team, so a client read the report because you forwarded it. Any address can be added from the recipient picker now, by one of two routes.

• Without a sending domain of your own, we email that person a link to confirm. Until they click it they receive nothing.
• On a verified sending domain you tick a box confirming they have agreed to receive your reports, and the address is added straight away with no email sent to them.

You can take someone off your report address book

There was no way to remove a recipient once they were in the address book, short of deleting the report that reached them. Anyone in the book can be removed now, and an address we have stopped sending to says why: unsubscribed, mail bounced, or marked as spam.

Visual change alert rules can be written again

Writing new alert rules for Visual Changes was switched off while change detection was retuned, and that section of Alert rules sat greyed out with a note saying so. It takes new rules again, and the note has gone from the page and from the welcome email.

The first weekly report now arrives when you were told it would

A new account was told its first report would arrive about a day after signing up. It could not: a weekly report covers the last completed Monday to Sunday week, so one fired the day after signup covered a week that ended before the account existed, and it sent nothing at all. The weekly now stays on Monday, so the first one lands on the Monday after you sign up and covers the days since, alongside the monthly roundup on the 1st.

Signup stopped turning away a good email address

Signup checks that your email domain can receive mail at all, and a check that timed out counted as a failed check, so a valid address was refused with a message saying its domain cannot receive mail. A check that could not be made is no longer read as an answer: only a domain that definitively publishes no mail records is turned away.

Alerts that should have reached you and did not now do

Three separate faults were swallowing alerts before they were sent. All three are fixed, so damage that was found and then quietly dropped is reported on the day it is found.

• One family failing took the rest with it. A database stumble while the visual comparison was being looked up cost that run its Page Scores and Content Health alerts too, for every site in it. Each family now stands on its own.
• A run that had to be stopped and restarted counted as a run you were shown. Damage first seen in one was treated as already reported and stayed silent from then on.
• A break the day after a change nobody could judge, or one judged deliberate, was read as something you had already been told about. It is sent now.
• One deliberate trade-off: if you have a visual change rule set below 5%, a page that changes a little every day will alert on each of those days. Carrying the previous day's number forward instead would have silenced a genuine break the day after a small wobble, which is the worse failure.

Page Scores alerts now say whether it was mobile or desktop

Every monitored page is scored twice, once on mobile and once on desktop, and an alert that named neither made a desktop alert and a later mobile alert read as the same message sent twice. The device is now named beside the score in the email, in Slack and in alert history: "performance (Mobile)" rather than "performance".

• The two devices are judged separately. A mobile score sitting still below your threshold no longer sends a fresh alert every day because the desktop score moved, and a real mobile drop is no longer taken for one you had already been told about.
• A day the audit itself failed no longer counts as a score you were shown, so the alert that used to repeat the day after one has stopped.

Visual history now shows the capture a change was actually measured against

The "Previous" image beside a change was not always the one the change was measured against: a page is compared with the last good capture, normally the previous day's, and history showed the capture immediately before whether or not that was it. It now shows the one that was really used, dated, with a line naming why when the two differ.

• Side by side, the two images scroll, zoom and pan together and stay lined up.
• A comparison that could not load no longer sits on a spinner that never ends.

An action that fails now says so instead of looking like it worked

Several things could fail quietly and leave the screen looking as though they had worked: saving account settings, downloading your data, "Email me this report", the contact form and switching between sites. Each of them reports the failure now, and the marketing emails switch snaps back when its save does not go through.

• A message about waiting names the real wait. Asking for a report by email again says to try again tomorrow, rather than telling you to wait a minute.
• An expired session returns you to siteletter.com instead of leaving a dead screen behind.

"Resend email" works, and signup stopped blaming the wrong thing

Pressing "Resend email" on the welcome page did nothing at all, and it sends now, with a 60 second countdown before it can be pressed again. The verify and handover pages have also stopped calling your link used up when the real problem was a dropped connection or a server error, and they offer a retry.

• A website address that could not be used was only found after the form had been submitted, and the message appeared under the email box. It is checked before the form is sent, and the error now sits under the box it is about.
• An address that is already verified gets a login link instead of an error, and the form no longer gives away whether an address has an account.
• One signup email a minute per address, five an hour and ten a day. Correcting a typo and pressing resend straight away will be refused until the minute is up.

A guide on what belongs in a maintenance plan

New in Guides: what a care plan covers and what to leave out, with a scope sheet you can work through section by section, and how to write the scope into the agreement.

Logging in before verifying your email no longer leaves an empty account

If you signed up and then pressed "Log in" instead of clicking the verification link, you reached an empty account where adding a site failed, and clicking the verification link afterwards only told you to log in. Logging in now sets the account up the same way that link does: the credits every new account starts with, your site added, and its first scan started.

• An account already left empty this way is set up the next time you log in.
• Your agency is named after your site's domain, because only the verification page asks for an agency name. You can rename it in Settings.
• If a team has invited you, your own account is not set up while that invitation is open, so you can still accept it.

A domain with no published expiry date no longer reads as expiring today

Some country registries, .de, .nl, .eu, .at and .au among them, do not publish domain expiry dates at all, and the site overview was filling that gap with a large 0 above "days remaining". Those sites now read "Not published", with a sentence naming the registry and pointing at the registrar, which is where the renewal date actually lives.

• A lookup that genuinely failed now says what failed, and that the next daily scan will try again. A registry that can never return a date says so instead, so you are not left waiting on a check that cannot succeed.
• Where the registry publishes a registration date, the card shows that in place of an empty Expires row, and the progress bar with nothing to count down has gone.
• This wording is on the dashboard today. A report you share with a client still shows "Couldn't verify" and a dash for these domains, which was never wrong, only less than the dashboard now tells you.
• Czech .cz and Brazilian .br domains are not in this group. Those registries do publish a date, in a format the check cannot read today, so those sites still say the next scan will try again.
• Each site picks up the new wording at its next daily scan.

Working links no longer read as broken on firewall-protected sites

Some sites, government and enterprise hosts especially, turn away the lightweight request used to test a link and answer with an error, and that error read as a dead link on the dashboard, in emailed reports, in PDF exports and on shared report links. Anything about to be called broken is now requested again in full and that second answer is the one reported, so on an affected site the broken count drops and Content Health can turn from red or amber back to green.

The sidebar separates a site from the monitors you buy for it

Overview and Pages stay under Site, and the five monitors you switch on per site now sit under their own Monitors heading: Visual Changes, Uptime, Page Scores, Renewals and Content Health. Every page is the same page at the same address, so anything you have bookmarked still opens.

A site with a problem no longer sorts to the bottom of your portfolio

When a site was not being charged for its monitors, because they had only just been switched on or because the balance had run out, the portfolio table left its columns blank and ranked that site as though nothing was wrong. What was measured is shown again, so a broken page or an expiring certificate carries the site to the top of the list with its tile open.

• Page Scores, visual changes, broken links and images, and certificate status all show their last reading in its true colour, and hovering says the reading will not update until that monitor runs again.
• Hover text is straighter: a check that was refused, usually by bot protection, reads "couldn't verify" instead of being called stopped, and a site with only some of its monitors stopped no longer claims a reason for it.

Receipts for your top-ups now open

The billing page has a Receipts and invoices button, and until today it failed for everyone who pressed it. It now opens your billing portal at Stripe with an invoice for every top-up behind it.

• If you have bought credits already, the button starts working from your next purchase, and receipts for anything bought before that may not be listed there.
• A new account that has bought nothing yet gets a plain line saying there are no payments to show, instead of an error.
• The button used to say "Receipts and payment methods". No card is ever kept on file, so the name now says what is actually behind it.

Monitoring is prepaid credits now, and the monthly subscription is gone

There is no monthly subscription and no trial any more. You buy credits once, in a single payment, and each site spends from that balance every day for whichever of its five monitors are switched on, so a brochure site that runs one of them costs less than a shop that runs all five. Nothing renews and nothing is charged automatically, and what each monitor costs is on the pricing page.

• One payment, not a plan. A top-up is a single purchase, there is no monthly bill, and the credits sit on your balance until your sites spend them.
• You pay per site, per day, for what that site actually runs. Visual Changes, Page Scores, Content Health, Uptime and Renewals are switched on one site at a time, and switching one off stops that charge from the next day.
• The balance is what keeps monitoring running. If it reaches zero the checks stop, which is why the dashboard now shows how many days of monitoring you have left and emails you before you get there.
• A new account starts with a grant of credits instead of a trial, enough to run one site with every monitor on for about two weeks, and no card is needed to use it.
• Credits do not expire while your account is open, and unspent credits from a purchase can be refunded within 14 days of that purchase.

You can see how many days of monitoring are left

The sidebar now carries a permanent line showing days of monitoring left at the setup you have now, and a banner appears from three days down. Running out is the one failure that leaves a hole in a client's history, so there are warnings in front of it.

• A warning email at seven days, three days and one day left, to the account owner and to managers.
• If credits do run out, a second email names every site that stopped and when it stopped, and says plainly that no data exists for the gap.
• The trial-ending, payment-failed and subscription emails are gone with the subscription that sent them.

Each site has its own panel for its five monitors

Visual Changes, Page Scores, Content Health, Uptime and Renewals are switched on and off one site at a time, from a panel on the site itself. It states what each switch does to the money before you save, and if the balance will not cover everything you ticked it turns on what it can and names the rest rather than refusing the whole save.

Adding sites is one flow, and it takes a list

A single "Add sites" flow replaces the separate doors for adding a client and adding a domain. Paste a list, set the client and the monitors row by row, and create the lot in one go.

• Up to 50 domains a paste, separated however they came out of your spreadsheet.
• A row that cannot be used comes back marked and the good rows are still created, so one typo does not cost you the batch.
• One organisation holds up to 100 sites.

A report shows only the monitors that site paid for

When the same domain was monitored on more than one account, a section either account paid for could appear in both of their reports. A report now carries only the sections that were paid for on that site, over the period it covers.

Report sections now carry the monitor names

Site Status is Renewals, Broken Links & Images is Content Health, and Lighthouse is Page Scores, in the emailed report, the PDF and the shared link. Alerts use the same names, so one check is never called two things in front of a client.

The shared report link now reads like the emailed report

Same narrow sheet, your agency colour along a top bar rather than on the wordmark, and real score rings in place of flat tiles. It also prints the note about links and images that could not be checked, which the email already carried and the shared page did not, so the two documents stop disagreeing in front of a client.

Uptime says what it measured, and what it did not

Uptime now has more than one answer. A window nobody watched, or one the site refused, no longer prints a percentage at all.

• A period with no checks reads "No data". A period where the site turned the checks away reads "Couldn't verify".
• A period only partly covered prints its figure together with the span it is based on.
• Uptime alerts name the site instead of counting pages, because the uptime check runs on the homepage. The per-page uptime table has gone from the dashboard for the same reason.

Managers can now do almost everything an owner can

Billing, branding, adding sites, switching monitors and managing the team are all open to a manager. Transferring ownership is not, and a manager cannot create an owner or demote one.

An invitation works for someone who has no account yet

An invited teammate without a SiteLetter account can sign up from the emailed link, and one who already has an account signs in first and lands back on the invitation. Accepting moves that person into your organisation, so a sole owner has to hand ownership over or delete their own organisation before they can accept one.

Deleting a site or a client now closes its shared report links

A shared report link kept working after its site was deleted, so anyone still holding the address could open it. Deleting revokes those links now, which also means an old link you sent a client will stop opening.

Sites you already monitored came with credits

Every account that was monitoring sites before the switch to credits has been given a one-off grant for each site, enough for about four weeks with every monitor on and longer on a site running fewer. There is nothing to claim; the balance already shows it.

Scanning a site on demand is switched off for the moment

The "Scan now" button, which ran an extra scan outside the daily schedule, is temporarily switched off and will be back. Every site keeps its daily scan, and reports, alerts and schedules are all unchanged.

Pay for what you switch on, and nothing recurring

The monthly subscription is gone. You now top up a balance and each site spends from it once a day for whichever of three services it has turned on, so a brochure site that only needs the daily checks costs less than a shop that needs everything.

• Three services per site: Core monitoring, Uptime, and Renewals. Turn any of them on or off per site, and the charge stops the next day.
• Nothing renews and no card is kept on file. New accounts start with two weeks of full monitoring on one site, free.
• Unspent credits from a purchase are refundable for 14 days, and credits do not expire while your account is open.

Reports no longer claim more than was measured

Where a report could not measure something, it used to fill the gap with a good result. It now says what it could and could not cover, which is the difference between a report your client can trust and one that quietly flatters the month.

• A week that was only partly measured no longer reads as 100% uptime. The report states how much of the period was actually covered.
• Pages that were never compared are no longer folded into a no-visual-changes result. They are listed as not compared.
• A certificate that browsers reject no longer shows a green count of days remaining.
• An expiry date labelled as a renewal read like good news, so a certificate days from lapsing looked like one that had just been taken care of. An expiry date is now called an expiry date, in the report email, the PDF and the shared report link.

A recipient who cannot be emailed now says so

An address blocked after earlier bounces or spam reports still ended on a cheerful confirmation screen when it was added to a report, even though nothing sent to it would ever arrive. It now says the address cannot be delivered to, so you find that out before your client asks where the report went.

Reports and alerts arrive in a new layout

Every email the product sends has been redrawn: a cleaner frame, your agency colour running along the top, and no block of raw link text at the bottom for people to copy and paste.

Signing in no longer needs a second click

The sign-in link in your email now opens the dashboard directly instead of a page with a confirm button on it. If a mailbox security scanner opens the link before you get to it, the button comes back automatically for the next 90 days, with nothing to ask for or switch on.

Report history and alert history now cover every client at once

Both pages have moved out of the single-client view into the agency section of the sidebar, and list everything sent across your whole portfolio. Filter by client and by site, and the filter stays in the address bar, so a filtered view can be bookmarked or sent to someone.

Dropdowns open at full size again

On any page that loads its data first, a dropdown could open as a sliver a few pixels tall and could not be used. They open properly now. Empty lists were tidied up in the same pass, so a filter that matches nothing no longer claims nothing was ever sent for that client.

Every company that handles your data is now listed on one page

When a client's privacy review asks who else touches their data, there is now a sub-processors page to hand over, naming each company and what it does. The privacy policy, terms and cookie notice were revised alongside it.

A failed send no longer resends the whole report

If one address on a report's recipient list couldn't be reached, the send stopped there and the schedule treated the report as never sent, so it went out again the next hour to everyone who already had it. The send now continues past a failed address, that recipient is listed as not delivered, and the report goes out once.

Uptime alerts name the site in the subject line

A down alert used to arrive as "1 site is DOWN", so you had to open it to find out which one. When the alert covers a single site it now reads "example.com is DOWN", which is enough to act on from a lock screen. Batches covering several sites still show the count.

The unsubscribe link in emails goes somewhere

The unsubscribe link at the bottom of every email pointed at a page that was never built, so clicking it returned a 404. It now opens a confirmation page that removes the address for real. Links from older emails reach that page too, and where the link predates the change it explains how to get the address removed.

Visual checks stay honest when a page stays broken

Two fixes to how a page is compared against its previous capture.

• A page found broken used to become the new reference, so the next check compared broken against broken, found no difference, and went quiet about damage that was still there. Damaged captures are no longer used as the reference.
• When the reference image had aged out of storage, the page was treated as never scanned, which reads as no change. It now falls back to an older capture, and if there isn't one the page is marked unverified rather than clear.

Billing updates can't be lost in transit

A billing update from Stripe that failed to record the moment it arrived was treated as handled and never sent again, which could leave an account looking like an expired trial while it was being billed normally. Those updates are now retried until they're recorded.

Alert emails open the site they're about

Clicking through from an alert could land on whichever site you last had open, not the one the email named. Every alert now links straight to that site's overview, so an alert about one site can't show you another one's data.

One broken link is one problem, not one per page

A single broken link in a footer used to be counted once for every page it appeared on, so a small problem could read like a big one. Each broken link or image is now counted once, with the number of pages it reaches shown beside it.

Links that can't be checked aren't reported as broken

Some sites block automated checks on outbound links, and those were reported red alongside genuinely broken ones. They're now listed separately as unverified, so the broken count only holds things that are actually broken.

Long-standing problems no longer get reported as new

When a scan could only reach some of a site's pages, the next full scan compared against it and treated everything it found as brand new, including problems that had been there for weeks. Comparisons now only run against pages the previous scan actually opened.

The same domain under two clients reads as two sites

If you monitored one domain for two different clients, the portfolio built both rows from the same record, so pausing one silenced the other. Each client's row now stands on its own.

Find where a broken link or image sits on the page

Content Health findings now have a locate button. It opens that page's latest screenshot with the item outlined, so you can see whether a broken link is in the main navigation or buried in a footer nobody reads, without opening the site.

Healthy images no longer reported as broken

Images that render perfectly could show up as broken when a site serves different versions of the same image to different screens. The check now reads the version the browser actually loads, which clears a whole class of false reports on sites that have been migrated between hosts.

The overview lists every site worst-first

Your whole portfolio is now one table, sorted so the sites needing attention are the first rows you read. Paused sites sit at the bottom, and every site shows its own favicon in the overview and the pickers so you can find a client by sight.

Fewer false downtime alerts

Three separate causes of a site being reported down while it was serving visitors normally have been fixed. Incidents now also record what actually failed, so a certificate problem, a DNS problem and a timeout no longer all read the same.

• A site that turns away checks from one location no longer counts as confirming an outage. When two locations disagree, or one is refused, a third is asked before any alert goes out.
• Sites that send unusually large response headers, which some WordPress performance plugins produce, are now read correctly instead of being reported as unreachable.
• A shared-hosting security rule was rejecting the check before it reached the site. Those checks now go through.

Site cards now reflect what was actually checked

A site whose scan stopped part way used to read "Awaiting first scan" even though it had real performance, visual and uptime results. It now shows what was collected, broken assets only report "Clean" once assets have genuinely been checked, and icon addresses that some site builders serve like pages no longer take up one of your monitored page slots.

Settings now cascade from your agency down to each site

Set something once at agency level and every client and every site inherits it, including the ones you add later. Override a single client when that one needs to be different, and set alert rules and mute per site.

• Alert rules cascade across agency, client and site. Report templates are set once at agency level and cover every client and every site; who receives each one and when is set per site.
• Pause everything at once, a single client, or a single site. Scans, uptime checks, alerts and reports stop, and all your history stays.
• Report and alert recipients are now one agency-wide list instead of one per client, with your own team on it automatically and a heads-up when a report cannot reach someone. Adding an address outside your team is not in the dashboard yet, so a client still reads the report when you forward it.

Billing shows your exact next charge

The billing page now reads straight from Stripe: your slot count, any discount on your plan, and the exact amount and date of the next charge. You can also type a slot count directly instead of clicking one at a time.

Scheduled report delivery fixes

Fixed an issue that could stop scheduled reports from going out, and made delivery self-healing: a missed send now goes out later the same day instead of waiting for the next period.

• "Send now" no longer delays your next scheduled report.
• The next-send time on the Reports page now follows the client's timezone.
• New clients start with weekly and monthly report schedules already set up.
• If a report has no confirmed recipients, you now get a heads-up email instead of silence.

Scans now show live progress while they run

A progress bar appears whenever a scan is running, covering the first scan after signup, a manual "Scan now", and routine scheduled scans. It fills as each page is captured and audited, and flags a scan that has stopped making progress instead of leaving you guessing.

Switching between clients and sites now happens in one place

The rows of client and site tabs across the top of the dashboard have been replaced by a single searchable picker in the sidebar. The client and site you are currently viewing are always named there, so it is clear what the page is showing.

Signup no longer stalls on slow-responding sites

Adding a site whose sitemap or domain records were slow to answer could use up the entire signup request, leaving the account without a scheduled first scan or report. Page discovery now works to a time budget, and the certificate and domain readings are filled in by the first scheduled scan instead.

Client reports now pick up your agency brand color automatically

The report builder, live preview, and branding panel now use your organization's brand color by default, so a new client's reports match your agency without any extra setup. Leave a client's color blank to inherit it, or set one to override just for that client.

A pricing guide for your care plans

New in Guides: how much to charge for website maintenance, with real published rates across five regions and a calculator to put a number on your own client list.

Shared report links now show the full report

When you share a client's report, the link opens the complete branded report in one page: Performance, uptime, visual changes, broken links, SSL, and domain. The shared page stays private and isn't indexed by search engines.

A free SSL/HTTPS checker, no signup

Check any domain's certificate: whether it's valid, who issued it, and how many days until it expires. Handy for a quick look at a client's site, or a prospect's before you pitch.

Sharper SSL certificate checks

SSL monitoring now flags more certificate problems: expired, self-signed, untrusted, or issued for the wrong domain, not just certificates nearing expiry.

Ongoing issues no longer trigger a repeat alert on every scan.

When a check crosses one of your alert thresholds, SiteLetter now alerts you once and stays quiet while it stays over that threshold, instead of re-sending the same alert each time the site is checked. No more duplicate emails or Slack messages for a problem you already know about.

Checks blocked by a site's bot protection no longer look like failures.

When a site's security or bot protection blocks a check, SiteLetter now marks it "Couldn't verify" instead of counting it as broken or sending a false expiry alert. The dashboard and the weekly email show the same neutral state, so a blocked check never reads as a real problem on a client's report.

Page discovery now follows your robots.txt rules.

SiteLetter respects the crawl rules in your robots.txt when choosing which pages to monitor, and the pages view shows how many pages those rules keep out of monitoring, with a sample of the paths. Your homepage is always kept, and discovery holds up better on large sites with sprawling sitemaps.

Alternative comparisons now use more precise data.

The /alternatives matrix and most comparison pages were re-verified on this date. Each page carries its own 'last verified' stamp.

Captures and change tracking are steadier on busy sites.

A few patterns that used to throw off a scan no longer do. Newsletter popups and cookie banners are hidden before capture, rotating banners are frozen, and lazy-loaded images are forced to load, so unchanged pages stop reading as changed.

Screenshots and visual change detection both got sharper.

Page screenshots capture more reliably on complex layouts, with fewer missing chunks and less duplicated content. And when something on a page changes, the diff highlights just the area that moved instead of marking the whole page as different.

Mixed content and insecure links now show up on content health.

Mixed content flags http:// scripts, stylesheets, iframes and images loaded by an https page (browsers block or warn on these). Insecure links lists http:// anchor URLs. Both show as separate sections on content health, distinct from broken resources.