We run our own site on Arclo. Here is the scoreboard.

Search

Speed and conversion

Change log

Every change shipped to arclo.pro, newest first, with what we expected it to do. Outcomes are filled in after 7, 14 and 28 days from the same data as the scoreboard. A change with no outcome yet is still in its window.

1. Sep 22, 2026OtherPR #311, #312

   This page stops implying it has revenue: the accounts and scans cards say they count Arclo's own testing too, and the weekly release email drops a discontinued product name and a link to a host with no DNS record

   Expected: Honesty, not traffic. "Accounts created" called itself the number that pays the bills; it is an unfiltered row count, both rows are internal, and Arclo has no paying customers. No metric should move — the number on this page just stops being read as something it is not.

   Result: in its measurement window

   /proof · /api/proof

2. Sep 22, 2026FunnelPR #309

   Build-a-site CTA beside the scan box on the homepage, and eight brick-and-mortar practice pages with scan types of their own

   Expected: A visitor who wants a site rather than a report gets a door of their own: mockups built per homepage session rise over 28 days without scan starts falling. The eight new /for pages are judged on indexing first — impressions within 28 days of being indexed.

   Result: in its measurement window

   / · /for/*

3. Sep 22, 2026FunnelPR #303, #304

   A paying customer is returned to their own dashboard with a confirmation instead of the marketing homepage

   Expected: Stripe's success_url pointed at /api/billing/success, which is not a serverless function, so the SPA catch-all dropped every customer who had just paid onto the marketing homepage with no acknowledgement. Provisioning was never affected — that runs in the webhook. Judged on whether anyone has to ask whether their payment went through, not on any search metric.

   Result: in its measurement window

   /app/overview

4. Sep 22, 2026TechnicalPR #300

   Three routes the site links were answering Google with Vercel's SPA shell: /websites, /results and /tools/website-generator had no prerendered HTML

   Expected: /tools/website-generator is linked from the homepage and from all 36 marketing footers and was serving a 5,498-byte shell with no canonical. All three should move from unknown or "crawled – currently not indexed" to indexed within 28 days. Verified live after merge: 95,825 / 51,717 / 43,714 bytes, and /results now carries its noindex,follow in the served HTML.

   Result: in its measurement window

   /websites · /results · /tools/website-generator

5. Sep 22, 2026FunnelPR #301

   Landing on /scan puts the cursor in the scan box instead of leaving the visitor to find it

   Expected: Scan starts per /scan session over 28 days. /scan only became real HTML on Sept 20 (#284), so this is the first window in which the route has both a prerendered page and a focused input.

   Result: in its measurement window

   /scan

6. Sep 22, 2026TechnicalPR #296, #297, #299, #306, #307, #308

   Scan report and app trust: the screen that asks for the money stops selling an SEO tool, content and local SEO are no longer scored on sites where neither was measured, keyword seeding is tied to the practice's own city, the engineering commit log is no longer published to prospects, and Delete works on every CRM lead record

   Expected: Reporting integrity, not arclo.pro traffic. Report-to-mockup rate should hold or rise; no organic metric expected to move.

   Result: in its measurement window

   /report/free/* · /release-notes · /app/*

7. Sep 22, 2026TechnicalPR #298

   A refused Google sign-in now says why instead of failing silently

   Expected: Measurement, not traffic: signups lost at the Google step left no record, so the size of that leak was unknowable. No metric expected to move; the loop gains a number it did not have.

   Result: in its measurement window

   /login · /signup

8. Sep 21, 2026SEO contentPR #291

   Sitemap lastmod was telling Google four rewritten pages had not changed since April

   Expected: Google recrawls the four rewritten pages within 14 days, and /multi-location — which had slipped to "crawled – currently not indexed" — is re-indexed within 28.

   Result: in its measurement window

   /sitemap.xml · /use-cases · /multi-location · /results · /managed-site

9. Sep 21, 2026FunnelPR #292

   Every blog post ends with the managed-site offer and a link to the reader's own practice page

   Expected: Blog sessions stop dead-ending: scan starts per blog session rise over 28 days, and the new internal links give the /for pages a second crawl path.

   Result: in its measurement window

   /blog/*

10. Sep 21, 2026FunnelPR #290

    The claim flow stops rejecting the claim Arclo's own link sends you to, and says why when it fails

    Expected: Mockup-to-account rate over 28 days. The previous behaviour was a dead end reached from Arclo's own email.

    Result: in its measurement window

    /claim/*

11. Sep 21, 2026TechnicalPR #294, #295

    Every published email address moved off arclo.io to contact@arclo.pro; security headers on every route; prerender stops injecting the same assets twice

    Expected: The site printed addresses on a domain Arclo does not receive mail on, so a prospect who wrote in reached nobody. Judged on whether inbound mail arrives at all, not on any search metric.

    Result: in its measurement window

    /* · /contact

12. Sep 21, 2026OtherPR #293

    One internal ops digest a day replaces the per-event blast that was producing roughly 25 emails a day and burying the one that mattered

    Expected: Internal only. No arclo.pro metric expected to move.

    Result: in its measurement window

    /api/cron/daily-leads-digest

13. Sep 20, 2026FunnelPR #284

    /scan prerendered — every "Run a free scan" call to action on the site was landing on an empty SPA shell

    Expected: The route named by the site's main CTA returned no content to Google or to a slow connection. Scan starts per session over 28 days, and /scan indexed within 28 days.

    Result: in its measurement window

    /scan

14. Sep 20, 2026TechnicalPR #285, #286, #289

    Core Web Vitals: a measured zero is no longer thrown away, the CrUX fallback works, field and lab readings are stored separately, and two dashboard numbers that disagreed with their own sources were fixed

    Expected: Reporting integrity. The PageSpeed figure on this page is a lab reading and is now stored as one, so the daily series is comparable day to day. No metric expected to move.

    Result: in its measurement window

    /proof · /app/*

15. Sep 20, 2026FunnelPR #287, #288

    An existing customer can claim a site, the dead claim page is gone, and the claim screen offers annual or monthly

    Expected: Mockup-to-account rate over 28 days.

    Result: in its measurement window

    /claim/*

16. Sep 19, 2026TechnicalPR #278, #279, #280, #281, #282, #283

    Platform: the Neon compute burn stopped (no sub-hourly crons), per-request DDL removed, Run and Grow added as real purchasable plans charged at site approval, CI green again

    Expected: Platform work, not arclo.pro traffic. The pricing page gains two plans that can actually be bought; no organic metric expected to move.

    Result: in its measurement window

    /pricing · /api/*

17. Sep 18, 2026SpeedPR #277

    Homepage hero text no longer fades in from invisible: the headline and sub-copy are the Largest Contentful Paint element, and the fade kept them uncounted until it had run

    Expected: Stored PageSpeed mobile series (68 on Sept 17) rises above 75 within three readings after merge and LCP falls by at least 1s from 5.3s. Measured before merge in headless Chrome on slow 4G with 4x CPU: LCP 2.8–3.2s with the fade, 1.6–1.9s (equal to first paint) without. No organic or funnel metric expected to move on its own.

    Result: in its measurement window

    /

18. Sep 17, 2026FunnelPR #267

    Free scan understands all 18 practice types: dermatology, optometry, veterinary, urgent care and the other second-wave verticals get their own keyword set instead of falling into generic "Healthcare / Medical"

    Expected: A practice arriving from its /for/ page gets a report about its own searches, so report-to-mockup rate for those verticals matches the therapy and psychiatry rate instead of trailing it. Judged on the funnel step rates over 28 days.

    Result: in its measurement window

    / · /for/* · /report/free/*

19. Sep 17, 2026OtherPR #243

    Cold outreach engine: a list of local practices is scanned, given a built site, and sent a three-email sequence with CAN-SPAM footer, one-click unsubscribe, reply and bounce handling (platform; not yet sending)

    Expected: No arclo.pro page changed. Once DNS and SendGrid are set up, outreach scans and mockups will land in the funnel counts on this page. The scan call sends source=outreach but the scan table does not store it yet, so until that column exists outreach and organic scans cannot be told apart here — a trust item for the loop.

    Result: in its measurement window

    /api/cron/cold-outreach · /unsubscribe/*

20. Sep 17, 2026SpeedPR #276

    Prerendered pages stay on screen while their JavaScript loads instead of being replaced by a "Loading..." spinner and painted a second time

    Expected: PageSpeed mobile on the stored daily series goes from 67–68 back above 75 within three readings and LCP falls from 5.3s toward the first HTML paint (under 3s); Lighthouse put the homepage H1's render delay at 2.2s. No organic or funnel metric expected to move on its own.

    Result: in its measurement window

    / · /for/* · /vs/* · /*

21. Sep 16, 2026FunnelPR #271

    Google sign-in signups now counted in GA4 signup_completed (the account was created inside a server redirect, so no page was there to fire the event)

    Expected: Measurement, not traffic: once Google sign-in is live, GA4 signup_completed equals accounts created instead of undercounting by the Google share.

    Result: in its measurement window

    /signup · /login

22. Sep 16, 2026TechnicalPR #268, #269, #270, #272, #273, #274, #275

    App dashboard trust fixes: Technical SEO reads the real weekly crawl, Automation and Outreach stop failing and stop showing placeholder scores, SERP refresh finishes inside its budget and never records a timeout as "not ranking", publisher health is measured, Paid Search says when there is no Ads data, direct online bookings reach Reservations

    Expected: Reporting integrity for customers, not arclo.pro traffic: what the app shows is measured or says it is not. No metric expected to move.

    Result: in its measurement window

    /app/*

23. Sep 16, 2026SEO contentPR #265

    Homepage meta description cut from 253 to 157 characters so the snippet ends on the call to action instead of mid-sentence

    Expected: Homepage CTR on its brand impressions rises above 3% over the 28 days after merge (0.8% over the 28 days before, 2.9% over the last 7).

    Result: in its measurement window

    /

24. Sep 16, 2026TechnicalPR #264

    Share-preview image fixed: og:image pointed at a password-protected Vercel deployment host, so shared links had no picture

    Expected: Links to arclo.pro shared on LinkedIn, Slack, iMessage and X render with the image from now on; no organic-search metric expected to move.

    Result: in its measurement window

    /*

25. Sep 16, 2026FunnelPR #262, #263

    Scan result page: score and the three problems that matter shown before the gate; the site Arclo built unlocks with an email

    Expected: Report-to-email-capture and report-to-mockup rates go up because the prospect sees the score and their built site (blurred) without scrolling; judged on GA4 mockup_unlocked and mockup_built per scan_completed over 28 days.

    Result: in its measurement window

    /report/free/* · /preview/*

26. Sep 16, 2026SEO contentPR #260, #261

    Last four pages rewritten off the old SEO-tool copy (/use-cases, /multi-location, /results, /managed-site); hero pills and footer industries fit on phones

    Expected: /multi-location, which had slipped to "crawled - currently not indexed", is re-indexed within 28 days now that it describes the same offer as the rest of the site. No click change expected.

    Result: in its measurement window

    /use-cases · /multi-location · /results · /managed-site · /*

27. Sep 16, 2026TechnicalPR #256, #257, #258, #259

    Free scan scoring made trustworthy: canonical host resolved first, SERP timeouts no longer counted as "not ranking", median of three PageSpeed runs, low-confidence authority left out of the score, domain preflight fixed on Vercel

    Expected: Reporting integrity: prospects stop seeing a score that was mostly measurement error. Report-to-mockup rate should hold or rise; no organic-search metric expected to move.

    Result: in its measurement window

    /report/free/*

28. Sep 16, 2026SEO contentPR #254

    /hipaa page; "what you'd pay elsewhere" on pricing; plan buttons given equal weight

    Expected: /hipaa earns impressions on "hipaa compliant website for therapists"-type queries within 28 days of indexing; pricing-to-signup rate holds or rises.

    Result: in its measurement window

    /hipaa · /pricing

29. Sep 16, 2026FunnelPR #253

    Scan progress screen shows real findings as each step finishes, offers "email me when it's done", and refuses domains that don't resolve with a did-you-mean

    Expected: Fewer scans abandoned during the two-minute wait: scan_completed views per scan_started go up, and typo'd domains stop burning SERP credits.

    Result: in its measurement window

    / · /report/free/*

30. Sep 16, 2026TechnicalPR #252

    Sentinel and Hemingway stop inventing numbers: random rank changes, guessed volumes and simulated trends replaced with real history or null

    Expected: Reporting integrity, not traffic: what the dashboard and reports show is now measured or absent. No metric expected to move.

    Result: in its measurement window

    /api/sentinel/data · /api/hemingway/data

31. Sep 16, 2026DesignPR #247

    Homepage shows real sites Arclo runs; How It Works rewritten; /results unlisted until it has a named case study; mobile menu and scan modal cleaned up

    Expected: Scan starts per homepage visit go up now that the page shows real work instead of placeholders; judged on GA4 scan_started per session over 28 days.

    Result: in its measurement window

    / · /how-it-works · /results

32. Sep 16, 2026FunnelPR #255

    Google Analytics installed on arclo.pro with five funnel events (scan started, report viewed, mockup built, claim clicked, account created)

    Expected: Measurement, not traffic: the site had no analytics tag at all, so step-to-step funnel rates were unknowable. From here the loop judges CTA and design changes on these rates.

    Result: in its measurement window

    /*

33. Sep 16, 2026SEO contentPR #250

    Footer "Compare" column: every page now links to the comparison and roundup pages

    Expected: These five URLs go from unknown to Google to crawled within 14 days and indexed within 28; they had no site-wide internal link before.

    Result: in its measurement window

    /vs/brighter-vision · /vs/therapysites · /best-therapist-website-builders · /best-med-spa-website-builders · /psychology-today-vs-your-own-website

34. Sep 16, 2026TechnicalPR #249

    Core Web Vitals trend now comes from stored daily readings instead of the live score with random jitter

    Expected: Reporting integrity, not traffic: the trend chart in every report showed movement that never happened. No metric expected to move.

    Result: in its measurement window

    /proof · /api/crew/speedster/summary

35. Sep 16, 2026SEO contentPR #248

    Titles under 60 and meta descriptions under 160 characters on all 18 practice pages

    Expected: Once these pages are indexed, CTR on their impressions beats the site-wide brand-only 0.8% because Google stops truncating the snippet.

    Result: in its measurement window

    /for/*

36. Sep 16, 2026FunnelPR #237, #238

    Mockup preview always shows the claim CTA; scan modal scrolls on short screens

    Expected: Preview-to-claim rate goes up: the claim bar used to sit under a page-tall iframe and the scan modal's Start button was below the fold on laptops.

    Result: in its measurement window

    /preview/* · /

37. Sep 16, 2026SEO contentPR #233, #239

    Website-builder roundups for therapists and med spas; Psychology Today vs your own website

    Expected: First impressions on "best website builder for therapists" (~390/mo) and related queries within 28 days of indexing.

    Result: in its measurement window

    /best-therapist-website-builders · /best-med-spa-website-builders · /psychology-today-vs-your-own-website

38. Sep 16, 2026SEO contentPR #228

    Comparison pages: Arclo vs Brighter Vision, Arclo vs TherapySites

    Expected: Impressions on "brighter vision alternative" and similar comparison queries within 28 days of indexing.

    Result: in its measurement window

    /vs/brighter-vision · /vs/therapysites

39. Sep 16, 2026SEO contentPR #226, #232

    Eleven blog posts for practice owners; post data embedded so Google's render is not a soft 404

    Expected: First non-brand impressions within 28 days; blog URLs move from "unknown to Google" to indexed.

    Result: in its measurement window

    /blog · /blog/*

40. Sep 15, 2026SEO contentPR #225

    Eighteen practice-vertical landing pages (/for/mental-health … /for/audiology)

    Expected: Impressions on "<vertical> website" and "<vertical> marketing" queries within 28 days of indexing.

    Result: in its measurement window

    /for/*

41. Sep 15, 2026FunnelPR #219, #221, #223

    Free report de-duplicated; practice-first business-type picker; report CTA opens the site builder with the scan attached; logo and hero upload

    Expected: Report-to-mockup rate goes up because the builder no longer re-asks what the scan already knows.

    Result: in its measurement window

    /report/free/* · /tools/website-generator · /preview/*

42. Sep 15, 2026DesignPR #216, #218

    Log In button in the header on every page; free-report hero text and tab bar made readable

    Expected: Fewer returning customers bouncing to find login; no measurable organic effect expected.

    Result: in its measurement window

    /* · /report/free/*

43. Sep 14, 2026SEO contentPR #209, #210, #211, #212, #213

    Front door rewritten around the managed-site offer; Google Ads add-on page; MCP docs; reporting made trustworthy

    Expected: Homepage CTR on brand queries improves; scan starts per visit go up.

    Result: in its measurement window

    / · /ads · /docs/mcp

How to read this