It has been a busy few days. First Tutors announced their closure on 8 May, and that — together with the tutors and parents who have been writing to us — has shaped most of what we shipped this week. The summary below sets out where the work has gone.
The response from the community since launch has been remarkable. Thank you to everyone who has signed up, written to us with feedback, and in particular to those who have taken the time to leave a review on Trustpilot. It is the early traction we have been working towards, and it is what made it possible to ship this much in a single week.
The First Tutors closure
Within hours of the announcement we published a page for affected tutors carrying a verified copy-paste template for a data-access request — a letter you can send the administrators asking for your contact list and reviews back, citing the relevant UK GDPR provisions. We followed it with a step-by-step guide to extracting the email source for Gmail, Apple Mail, Thunderbird and Proton Mail, since the DKIM signature on the closure email is what makes the data-access claim watertight.
Trust and safety
A directory is only as good as the trust people place in it. The bulk of our engineering hours this week went into trust and safety, and it is worth setting out what changed because almost none of this work is visible day-to-day. That is by design; the platform should feel calm and quiet, but the protection underneath should be real.
The work this week breaks down into two surfaces: messages and reviews.
Messages. Our redactor strips contact details — phone numbers, emails, URLs, common chat handles — from any message until the unlock fee is paid. That has been in place since launch and works well per-message. What was missing was a way to spot the pattern of an account fanning out the same templated pitch to many tutors at once.
A recent case shaped the work. A new account signed up and, within the hour, sent near-identical pitches to roughly two dozen tutors — each one carrying a link to a phishing site disguised as something innocuous, such as a meeting URL. The redactor caught the link in every message, so nobody was able to click through. The harder problem was joining the dots: each tutor only saw their own message, with no way to know that twenty-three others had received the same pitch. We acted after one tutor reported it and suspended the account, but it took longer than we would have liked to recognise the pattern.
What shipped this week is built around exactly that shape of attack. We now look at each account's outgoing activity in aggregate, rather than one message at a time, so the next time someone fans out a templated pitch the account is held for review long before two dozen tutors have received it. We have deliberately chosen not to describe what specifically triggers a flag — the value of a system like this is that those who would abuse it cannot work around it. The threshold is set so that a genuine parent shopping around with a copy-pasted "looking for a GCSE maths tutor on Tuesdays" will not trip it.
Reviews. Review fraud is the other obvious shape to guard against on a directory — a tutor with a friend prepared to play a customer could in principle inflate their reputation. The same engine now examines incoming reviews and pauses anything that matches a handful of patterns associated with review fraud — again, without us being specific about the signals. Genuine reviews should pass through unaffected. Where something looks borderline, the review sits on an admin queue rather than going live; we look at it, ask the tutor for context if useful, and make the call.
A few smaller pieces sit alongside that work:
- Reviews can now only be submitted seven days after the finder's fee is paid. This rules out the same-day "I unlocked, then left a scathing review" pattern that some platforms have tolerated for years, and gives both sides time to try the introduction.
- The Report button is no longer hidden. It now sits in the thread header where it belongs, alongside an inline banner that appears the first time we redact contact details from a message you have received — with a link to a new scam-spotting guide.
- A reversible "Suspend" action replaces our old "Ban" admin tool. It locks the account, hides the tutor profile, and freezes their threads in a single step — and because the prior state is preserved, a false positive can be unwound cleanly. It lets us act decisively without leaving ourselves stuck if we have got something wrong.
- Messages from a suspended author are removed from the other side's view, so a friendly-looking exchange around a now-removed scam pitch does not read as legitimate after the fact.
A note on false positives. If the system locks an account that turns out to be a genuine parent or tutor, the lock is reviewed within 24 hours (slightly longer at weekends), the account is unlocked silently, and the user can carry on as before. The user-facing messaging is deliberately gentle — "your account is being reviewed" rather than "you have been flagged as a scammer" — because most cases that reach us are inconvenience, not malice. If it happens to you and you believe it is wrong, write to us and we will resolve it the same day.
Identity and safeguarding
Identity verification (the Stripe Identity check carried out as part of becoming a verified tutor) is usually a single-attempt process, but in rare cases — poor lighting, glare on the ID, an unclear selfie — a tutor can run through all three attempts without getting through. There is now a self-serve "Pay for three more attempts" button on that page, alongside refund and email-support options, so anyone in that position can continue without waiting for us. Administrators can also issue a free retry on a case-by-case basis. We now also send our own receipt when the verification fee is paid, rather than relying solely on Stripe's auto-receipt, which had previously been the only confirmation tutors received.
On safeguarding (DBS, PVG, AccessNI), we shipped the part of the workflow that handles what happens after a tutor is verified. Background checks do not last forever — every certificate carries an expiry date — and tracking that should be the platform's job, not the tutor's. Every verified tutor now sees a Safeguarding card on /me showing their current status and the date their check expires, receives a reminder email well before that date (with further reminders as it approaches, so it does not creep up on anyone), and — for tutors on the DBS Update Service — can renew their badge in a single click without re-uploading anything.
PVG (the Scottish equivalent of DBS) has now been tested end-to-end and is confirmed working, with thanks to the Scottish tutor who helped us get there.
Feedback we acted on
Tutors have been writing in steadily over the past few days, and we have tried to keep pace:
- The bio editor was rejecting framework names like
Vue.js,Node.jsandconfig.jsonas if they were URLs, blocking a software tutor from publishing an otherwise normal bio. The contact-detail filter now exempts known file-extension and language suffixes. - The marketing-consent box was pre-ticked at sign-up. Raised on Facebook as not being UK GDPR compliant, which is correct: it ought to have been opt-in from the outset. It is now unticked by default, and the privacy policy sets out the legal basis.
- The DBS approval process is now more robust — particularly for tutors with compound or non-English surnames, where the previous logic occasionally returned a no-match against an otherwise valid certificate.
- Stale browser tabs no longer cause silent errors. A tab left open across one of our deploys could occasionally cause a form to fail without explanation. Open tabs now detect the new version and reload on your next navigation.
- Email typos at sign-in are caught earlier. A common one (
@icloud.commistyped as@icloud.con) had been bouncing without explanation. We now suggest the likely correction on the first submission, and accept whatever you choose on the second. - Display-name UX has been tightened up. Sign-up now shows an
e.g. Jane Smithhint on the name field, and placeholder values such as "Your Name", "Test" and "Example" are rejected on submit — closing a case where an account had ended up literally named "Your Name". - BA2 postcodes now resolve to Bath rather than Wiltshire, fixing a specific outcode that had been pinning tutors in Bath to a town some distance away. A broader review of postcode-to-town accuracy is still in train.
- Scottish qualifications are better represented. MA (Hons) and MLitt are now in the degree dropdown, and the level labels read "Higher" and "Advanced Higher" rather than the plural forms — to match how a Scottish tutor would actually phrase it.
A handful of others — group vs 1-to-1 pricing, in-person teaching-location options, and a few subject-taxonomy gaps — are logged and on the list. We will work through them over the coming weeks.
Other changes
- Sending a chat message is now near-instant: the work that notifies the other party runs in the background rather than holding up the UI. Scroll behaviour has been fixed too — the conversation no longer over-scrolls and pushes early messages off screen.
- Cmd / Ctrl + Enter sends a message in the marketplace, support, and admin chat composers — the keyboard shortcut you probably tried first.
- Price high-to-low sort is now available alongside low-to-high on the directory pages.
- The tutors directory now shows the real total count in the toolbar. It had been reading "24 tutors" once supply crossed the first page, regardless of how many were actually listed; it now reads, e.g., "Showing 24 of 86 tutors".
- Geology is now available as a subject, and EPQ as both a subject and a level. Both were raised by a tutor on the same day.
- The footer and T&Cs now name Tutorperch Ltd, CRN 17207346, alongside the registered office. Our processors (Resend, Google Workspace) are now disclosed in the privacy notice.
- Search engines were occasionally indexing
http://versions of our pages due to how we built outbound URLs. That has been fixed at the source.
What's next
Over the next few weeks the priority is on two fronts: spreading the word, and continuing to help tutors recover their data from First Tutors. Alongside that, we are working on a code of conduct for the platform and a more rigorous set of controls on tutor profiles — both shaped by tutor feedback — together with greater flexibility in how tutors present their qualifications and subjects.
If you have spotted something — broken, missing, or simply awkward — please write to us. Feedback acted on within 24 hours is a far better outcome than feedback that waits a quarter, and we want to keep that pace.
— Robert & Fiona