POLYWOOD left a custom Magento build that took most of the engineering time
A vendor case study of POLYWOOD moving from Magento to Shopify.
Read the case studyWhat happened when people actually moved.
Each post is a case study. A team or person left one tool for another. We look at why, what they gained, what they hit, and the lesson you can reuse.
Four ShipFast domains left Mailgun for Resend and stopped forwarding replies
Bill Nalen moved four ShipFast domains off Mailgun in December 2023. Magic links stay on NextAuth. The mg subdomain goes away. In his write-up Resend does not forward replies, and each domain ends up with its own Resend account.
Read the case studyA Squarespace export imported posts as Published and left the images on the old host
Rob Pickering moved robpickering.com on 20 July 2010 with a per-journal Movable Type file. Posts imported as Published with dates, authors, and comments, while the images kept loading from Squarespace.
Read the case studyDEV moved production Postgres from Heroku to Neon with Bucardo
The DEV team kept the app on Heroku and moved the primary database after more than eight years. Bucardo ran on a DigitalOcean droplet. Leftover triggers forced a restart, and pooling took a few days. They disclose a Neon sponsorship on the same post.
Read the case studyGiFT's corporate site left Firebase Hosting for Vercel
Naoki Ioroi at GiFT Inc. published the cutover on 5 August 2026. Firebase Hosting had been rewriting every request to Cloud Functions v2, with Remix assets copied by hand. The same post also moves Contentful to microCMS. This site has a hosting playbook and no microCMS one.
Read the case studyhochzeitsplaza.de left Contentful for Strapi and built its own importer
A guest guide on Strapi's blog says a price rise made the move unavoidable for hochzeitsplaza.de. About 2,000 entries and over 11 GB of assets took about 7 hours. The tool is the author's, and the page prints no euro figure.
Read the case studyIndeed turned Jenkins off the day the CloudBees license expired
Carl Myers describes Indeed's CI move from Hudson in 2007 to a Jenkins shutdown on 1 April 2024. About 11 platform engineers, tens of thousands of repos, and over 2,000 Jenkinsfiles still on the old system when the date was set. GitLab CI now runs over 25,000 pipelines a day.
Read the case studyKuda left Fivetran credit pricing for Airbyte and BigQuery
Airbyte's 1 February 2026 case study quotes Kuda's head of BI on SQL Server CDC into BigQuery. Credits ran out mid-cycle. The cutover took under three weeks, and some tables needed 48 to 72 hours to backfill. No dollar amount is on the page.
Read the case studyMux moved a Ghost blog into Sanity so it could ship with the marketing site
Dylan Jhaveri wrote two Sanity pages on the same day. The blog post says about 130 Ghost posts. The import guide says 128, over a couple of months. Ghost allowed one author. The new post type allows several, and the same people power the team page.
Read the case studyToolBox left Vercel for a static export on Cloudflare Pages
Johin describes a Next.js tool site, 139-plus pages and 20 API routes, moving off Vercel on 3 March 2026. The first Cloudflare deploy failed a 25 MB Worker limit. He also says the project uses the OpenNext adapter, which is a different runtime story from a pure static export.
Read the case studyTraba spent a year moving Firestore collections to Postgres
Traba's engineering post walks a year of collection moves off Firestore onto Aiven Postgres with Prisma. Triggers plus reconciliation hit 99.99% precision in April. The shifts cutover flipped reads and writes together because replication lagged about a second. Firestore stayed for flags and cron.
Read the case studyZencity folded per-app Sentry into one Datadog shell
Shai Malul describes 10-plus micro-frontends, each with its own Sentry, moving frontend logs to Datadog RUM in the shell app. A module-federation SDK did not fit. About 200 logging calls in each of 5-plus repos were rewritten with JSCodeshift. Sentry is not described as uninstalled.
Read the case studyZenML rebuilt 2,224 Webflow pages in Astro during a one-week hackathon
Alex Strick van Linschoten moved zenml.io off Webflow in the February 2026 Claude Code hackathon. The page counts 2,224 generated pages, 2,151 crawled URLs, and 2,255 built HTML files as different sets. A five-day Webflow cache header kept the old site up after DNS moved.
Read the case studyAngular apps left Heroku the week the free dynos ended
A few Angular SSR apps moved from Heroku to Vercel in November 2022, the week free dynos ended. The published change wraps the Angular server build in one Vercel function. One build error cleared after Node was set to 14x. The post does not move a database.
Read the case studyBack Market retired Snowflake and Databricks after BigQuery was already the copy
Back Market's data already lived in BigQuery. The 2025 Google Cloud case study turned off Snowflake and Databricks and moved the wider stack off AWS. Earlier engineering posts show the Delta Lake landing, the Tableau check, and a later query-cost spike. The 90% figure is their CDC pipeline.
Read the case studyFrump Fighters left Zapier for n8n and got a $400 plan refund
Nuance moved Frump Fighters off Zapier in 2021. Corina got more than $400 back on the yearly plan. The $5 a month line is Nuance's server pitch, not her bill. The page never lists the zaps.
Read the case studyIngewikkeld imported Slack users and left a decade of messages behind
Ingewikkeld moved from Slack to Mattermost on Nexaa in early 2025. Users and channels imported. A decade of messages hit an upload limit and stayed in an offline archive. The server is a 1 CPU, 2 GB Team Edition container on MySQL.
Read the case studyMicrosoft moved 1,600 Azure DevOps repos to GitHub and kept Boards and Pipelines
Microsoft's CAP group moved repos to GitHub over six months and left Azure Boards and Pipelines in place. The same post says both 80% of in-scope repos and 70% of active in-scope repos. Enterprise Live Migrator is the tool they expect for the mono repos still left.
Read the case studyMidHard left Slack after a 30-day shutdown notice
Slack told MidHard Games on 19 December 2025 that the workspace would be suspended in 30 days. They exported with slackdump, imported into Mattermost on Yandex Cloud, and kept Slack for some partners. Their price table is rounded, and the license limits need Mattermost's own pages next to them.
Read the case studyNovu's Route 53 to Cloudflare cutover was down for 55 minutes
Novu moved DNS from Route 53 to Cloudflare on 23 March 2025. The outage ran 22:00 to 22:55 UTC for every user. Lower TTLs and cache flushes did not stop stale nameservers or negative caching. The postmortem gives no record count and no TTL values.
Read the case studyOutschool mirrored GitHub while GitLab became the place merges landed
Outschool imported GitHub into GitLab in 2022, then mirrored merge requests back so GitHub Actions and Heroku review apps kept running. A few senior engineers finished the GitLab move and the Heroku-to-Fargate move together in under three months. Traffic shifted on a Cloudflare load balancer over a few days.
Read the case studyStarburst's Pulumi page says Galaxy deploys fell from two weeks to three hours
Pulumi's Starburst case study quotes a two-week Terraform deploy and "about three hours a day" after. The same page also says three hours flat, 112x, and a 99% cut. A 2023 Pulumi blog about the same engineer says 3x and a few minutes. Those are different sentences.
Read the case studyTwo Route 53 zones moved to Cloudflare once tunnels were the reason to stay
Two personal domains left Route 53 for Cloudflare DNS on 13 January 2025. The bill was about 12 euros a year. The cutover took about 45 minutes, and Ansible then left stale records behind. A later post shows the tunnel and DNS-01 setup still in place.
Read the case studyLeaving Ghost Pro for Substack merged every list into one inbox
Candost left Ghost Pro for Substack so he could stop designing the site. Substack subscribed every address to every section, and his essay does not walk the importer.
Read the case studyGong routed non-urgent Datadog alerts into Sentry and left OpsGenie for pages
On 19 June 2022 Ran Mochary described a Datadog webhook bridge into Sentry. Urgent pages stayed on OpsGenie. A 10-minute beacon pages if the bridge goes quiet.
Read the case studyKDE Plasma moved the Ocean design system from Figma to Penpot
On 25 October 2025 Andy Betts wrote that Plasma's Ocean design system is leaving Figma for Penpot. Variants landed. The UI is not in Plasma yet, and the file still sits in a personal GitLab repo.
Read the case studyA media publisher moved about 50 million monthly searches off Algolia
Ksolves says an unnamed publisher cut search cost by 80% with self-hosted Typesense on AWS. The page is a consultancy case study, and the cluster is not Typesense Cloud.
Read the case studyA thousand LastPass items moved to Bitwarden after the vault breach
On 3 January 2023 David J Eddy imported 1,000-plus LastPass items into Bitwarden in a few minutes. PGP notes were too long for a password item, and folders could not become nested groups.
Read the case studyMixpanel Engineering left Medium and kept the old archive online
Neal Kaviratna and Austin Pray moved the Mixpanel engineering blog to Substack on 16 May 2025. The public post is a few paragraphs, and the old Medium URLs stay up.
Read the case studyRazroo left Pinecone namespaces for Qdrant filters and UUIDs
Charlie Greenman says Razroo moved off Pinecone on 4 April 2025. Qdrant filters replaced namespaces, string ids became UUIDs, and the speed claims are Razroo's.
Read the case studyRevelry copied thousands of Pinecone vectors into Qdrant in about three minutes
Revelry's Elixir app already had a module per vector database. On 20 March 2024 they copied thousands of Pinecone vectors to Qdrant in batches of 100, and they still support both.
Read the case studySpotflow left SendGrid for Resend after verification mail sat in Gmail
Michael Mikuš measured SendGrid, Postmark, and Resend with GlockApps, then moved Spotflow's Auth0 verification mail to Resend. A template rewrite removed a four-minute Gmail delay that the provider swap did not.
Read the case studySteamLevels left Mailgun, then SendGrid, for Postmark shared IPs
In December 2020, jfx wrote that SteamLevels outgrew Mailgun and a SendGrid spam pool before landing on Postmark. About 5,000 emails a month made a dedicated IP the wrong fix.
Read the case studyLeaving Airtable for Notion after features moved behind a higher plan
On 10 November 2023 the write-up said operations were moving from Airtable to Notion. The $450 a year is four tools together. This post has the reasons, and it does not have the field map.
Read the case studyChannel.io left CircleCI outages for self-hosted GitHub Actions
Jaehong Jung moved Channel.io's CI from CircleCI to GitHub Actions on ARC. Self-hosted runners cut CI cost to about a third. Karpenter, Docker Hub limits, and GitHub's cache cap showed up in production.
Read the case studyDual-running Matomo and Google Analytics for a month
Google Analytics was replaced with a self-hosted Matomo install that skips cookies and honors Do Not Track. A one-month overlap showed Matomo pageviews about 11 percent lower.
Read the case studyLeaving Google Analytics for self-hosted Matomo after Schrems II
On 20 January 2022 a blog replaced Google Analytics with Matomo and MariaDB in Docker. The snag was container networking. The write-up mentions a GA import and does not describe running one.
Read the case studyHelp Scout rebuilt a Sketch design system in Figma in a week
Buzz Usborne rebuilt Help Scout's Sketch libraries in Figma in a week. Nested symbols did not import cleanly, so he traced them. In-flight projects stayed in Sketch.
Read the case studyHubSpot left Mixpanel, built an internal tool, then moved to Amplitude
Dan Wolchonok's January 2018 post traces HubSpot from Mixpanel in 2013, to an internal product-analytics tool, to Amplitude. About 250 to 300 people a month used Amplitude after the 2017 cutover. He does not describe an event export.
Read the case studyIMG Licensing left Contentful after the free tier stopped fitting
John Kavanagh moved IMG Licensing from Contentful to Sanity in the same week he published, after the next Contentful plan crossed $300 a month. The front end was already on Next.js. He does not publish the import log.
Read the case studyMercari accepted broken Confluence imports and kept a closed-stack archive
Kiko Ando wrote up Mercari's Confluence to Notion project on 31 March 2026. The import was never perfect. Finance docs stayed out, and leftover pages are heading for a closed-stack archive.
Read the case studySemrush rebuilt Sketch libraries in Figma in just over a month
Five Semrush designers rebuilt the Sketch libraries in Figma between 19 March and 24 April 2020, alongside their normal work. Google Drive duplicates and Zeplin were the pain. Nested symbols were rebuilt.
Read the case studyThe Fork moved 4,500 Confluence pages in two weeks
Alegria.group says it moved more than 4,500 Confluence pages and 10 GB into Notion for The Fork in two weeks. The page is an agency case study. It names no Fork employee and lists no broken import.
Read the case studyBaserow left GitLab because the community lived on GitHub
Bram Wiepjes moved primary development in November 2025. Rebuilding CI was the long job. Issues and open merge requests went over with a modified node-gitlab-2-github. GitLab stayed a read-only mirror.
Read the case studyClear Street spent two and a half quarters moving GitLab to GitHub
Self-hosted GitLab since 2018. Part 1 is the CI pain. Part 2 is dual CI, a tool called Octomigrator, and a mirror that lagged up to 15 minutes.
Read the case study1Password Concierge credited unused LastPass time
The move was published on 7 March 2023. The import is the official desktop path. The credit depended on a payment-history PDF and a helper bot whose URL now 404s.
Read the case studyA hobby app left PlanetScale for Neon after the free plan ended
PlanetScale retired the Hobby plan on 8 April 2024. This 2024 write-up on mostviertel.tech uses pgloader, quote identifiers for Prisma names, and a baseline because PlanetScale migrations would not export.
Read the case studyA Notion export as HTML, then a Markdown cleanup
On 25 April 2025 he exported the workspace to HTML and ran Obsidian Importer. Large databases needed hand edits. Dataview lost to the Tasks plugin. The vault is the website repo, synced with Syncthing.
Read the case studyA month leaving Notion, and the web-clipper links got dropped
Notion went from $8 to $12 a month. Obsidian Sync was $4. He had tried twice before. The third try took a month, thousands of documents, and a written-off clipper archive.
Read the case studyA Notion to Obsidian import that dumped thousands of images in the vault root
The switch was 1 January 2025, after Notion since 2018. The Importer kept the Notion tree and left the pictures in one folder. Cleanup took a couple of months.
Read the case studySkyscrapers finished Terraform to OpenTofu 1.6.2 in four phases
Frederik Denkens says the license change from Terraform 1.6.0 pushed the move. By the end of May 2024 the company was on OpenTofu 1.6.2. The post names the phases and skips the state file.
Read the case studyTrueCar moved 700 Spacelift stacks off Terraform
Matt Velez wrote this as a guest post on Spacelift's blog. More than 700 stacks climbed Terraform 1.5.5, then OpenTofu 1.6.2, then 1.9.0. An S3 backend key built from locals makes Terraform init fail on purpose.
Read the case studyWhiteUI.Store left Gumroad after PayPal payouts stopped
The 4 November 2024 post quotes a Gumroad email about PayPal. Buyers keep old files on Gumroad. Lemon Squeezy's docs can import products, customers, and license keys, and they say subscriptions do not move.
Read the case studyButtondown replaced GitHub Issues with Linear and left the public roadmap on GitHub
Justin Duke announced the internal move on 30 October 2024. A 9 January 2025 stack note confirms Linear replaced GitHub Issues. Notable roadmap items stay on GitHub and sync into Linear. He never describes the importer.
Read the case studyButtondown replaced Statuspage and PagerDuty with Better Stack
Justin Duke's 9 January 2025 stack note. Better Stack, already their logging tool, replaced PagerDuty and Statuspage for alerting. The page is one sentence on that cutover, and it warns that the list may be stale.
Read the case studyCodePen left MySQL 5.6 for PostgreSQL instead of upgrading in place
CodePen Radio 351, 20 January 2022. Alex Vazquez and Chris Coyier describe a move to PostgreSQL 13.3 with about a minute of read-only mode. AWS DMS kept MySQL and PostgreSQL in sync. Elasticsearch was a reason, not a finished removal.
Read the case studyCoyote left HubSpot CRM for Salesforce after teams could not share a customer view
A Think RevOps interview with Oliver Durham-Winch, Coyote's global head of sales. OKRs came first. Then HubSpot CRM data moved to Salesforce, with HubSpot, Intercom, and Productboard connected. The 3 to 4 hours is his line.
Read the case studyLeaving Google Docs for Obsidian because Drive search missed the notes
A 16 June 2024 post, updated 10 June 2026. Technical notes moved to local Markdown, synced with the Obsidian Git plugin to a private GitHub repo. The page says to migrate only what you need, and it skips the exporter.
Read the case studyA cybersecurity firm kept HubSpot Marketing while moving CRM to Salesforce
Modelit's 2023 case study. An unnamed North American cybersecurity company left HubSpot's free CRM for Salesforce and kept Marketing Hub. Export limits and different tables drove the manual work. No record counts or dollar figures.
Read the case studyNylas left Jenkins because SOC 2 blocked third-party CI
Nylas published the move on 22 July 2020. A few weeks later, pull-request tests were 30% faster and some weeks had zero build failures. The title calls that a deployment-time cut.
Read the case studySlack used GitHub's importer plus an intern tool on 242 Jenkins pipelines
Emmanuel Zerefa's December 2024 intern post covers 242 pipelines on three Jenkins instances. The importer fully converted about half. The 1,300 hours saved is a model, not a finished measurement.
Read the case studyTeamdash moved about 200 tenant databases off MySQL for pgvector
Karl-Sander Erss, 28 September 2023. Teamdash left MySQL on Amazon RDS because vector search was missing. nmig moved about 200 databases in a four-hour window. Boolean columns mis-typed as integers showed up in production.
Read the case studyZealot dual-ran Amplitude and PostHog until Amplitude missed events
PostHog's 2 April 2025 interview with Zealot CTO Brandon Jakobson. They ran both analytics tools, then deleted Amplitude and BugSnag in one pull request. No historical import is described.
Read the case studyAxolo left Heroku for DigitalOcean App Platform and inherited double crons
Sydney Cohen moved a Node backend in two days. DNS took about 15 minutes on a Saturday. NODE_ENV stayed on Staging for a day, and EasyCron plus leftover Heroku jobs pinged users twice. The $560 versus $65 table is his forecast against the new bill.
Read the case studybalena shut down Jenkins after four weeks of GitHub Actions callers
The balenaOS team moved build, test, and deploy off a plugin-heavy Jenkins. Hardware-in-the-loop tests now take about one third of the time. External contributors can read the logs. This is a first-party post from 30 January 2025.
Read the case studyRepos moved to GitHub while SonarQube stayed bound to Bitbucket
Brian's team left Bitbucket for GitHub. The git remotes moved. SonarQube Cloud stayed on the old workspace until a GitHub owner rebound the org and every project.
Read the case studyA content team could not move a multi-year Trello board without losing comments
Sarah Yeung's team tried Trello's CSV and four other exporters. Comments, attachments, and subtasks stayed behind. Unito would not import history. Altosio came through with gaps.
Read the case studyEasy Eatery left SendGrid after shared IPs landed on block lists
Paul Dean's booking product saw Hotmail, Live, and Outlook rejects from SendGrid shared IPs. Less than 2% of mail failed. Detection to Postmark took under 48 hours. This post is thin on DNS.
Read the case studyFreeAgent moved a 58,000-test Rails suite off Jenkins
Daniel Holz wrote the why in June 2022. p95 deploys dropped from 25 minutes to 16. Test-infra spend fell about 90%. Part 2, the how, was never published. A later Dev Platform post confirms the move stuck.
Read the case studyA LastPass to 1Password move in 25 minutes after a broken export
After the LastPass breach, a CSV backup rendered most passwords in the browser. 1Password imported categories as tags in 25 minutes, then asked for the LastPass email, password, and 2FA.
Read the case studyLeaving LastPass after 10 years and missing notes search
After the 2022 breach he moved a decade of LastPass folders into 1Password vaults and tags. Archive and family sharing got better. Notes search, outside-family shares, and form-data save did not come with him.
Read the case studyMighty Networks members export cleanly and courses do not
A Circle consultant who has moved dozens of clients says Mighty Networks member data exports cleanly, courses need rebuilding, and a thorough move takes two to three weeks. This is a consultant summary, not one named company's dump.
Read the case studyA course site bounced from Elastic Email through SendGrid to Postmark in hours
A Teachable course was moving to WordPress. Shared IPs failed twice. Cloudflare DNS plus WP Mail SMTP landed on Postmark the same day. About 300 to 400 signups a day could not wait for a warmup.
Read the case studyCloudflare flattening is not a portable CNAME
He exported a BIND zone, dropped SOA and NS, and watched Route 53 reject the apex CNAME. Registrar transfer and a leftover DNSSEC key were the rest of the DNS move.
Read the case studyA finance platform left Chargebee and rebuilt checkout in Stripe Billing
DataArt moved an unnamed trading-indicators product off Chargebee. Customers and live subscription state went to Stripe. PayPal and Coinbase needed a custom checkout. This is a vendor case study.
Read the case studyDovetail Games moved 71,000 Freshdesk tickets into Zendesk
Help Desk Migration published a Freshdesk-to-Zendesk PDF for Dovetail Games. Tickets, contacts, and articles are listed as moved. A separate help-center write-up shows why article HTML is a different job. This is a vendor case study.
Read the case study3,481 Evernote notes exported after the official export failed
evernote-backup plus Obsidian Importer moved 28 notebooks. MFA needed OAuth. A rate limit paused the sync for half an hour. A 3,000-note notebook broke the desktop export.
Read the case studyLeaving Firebase for Appwrite without the password hashes
He moved a product off Firestore after lock-in and read costs started to hurt. The SDKs looked familiar. Auth still needed resets or a lazy migration, and Storage needed a custom script.
Read the case studyAn Intercom-to-Zendesk move succeeded and then lost snooze
Andy38's team moved support off Intercom. Ticket history can travel. Native snooze does not, and deferred work sat in the open queue.
Read the case studyPrime Time Palm Beach County replaced 140 in-person signatures
Paola Cedeno's team left DocuSign for Dropbox Sign so MOUs, RFPs, and W-9s could be templates with reminders. The Dropbox story does not describe a historical envelope import. This is a vendor case study.
Read the case studyA decade of Evernote notes needed a custom exporter before Obsidian
Peter Quiel left Evernote after the Linux client died. Official ENEX import capped out at 50 or 100 notes. He wrote an exporter, then used Obsidian's Importer.
Read the case studyA 1,000-note Evernote archive moved to Notion in minutes
週休3日サラリーマン used Notion's Evernote importer on a text-only vault. Five minutes of clicks, about ten minutes of waiting. Empty lines disappeared. Do not copy this timing onto attachments or huge notebooks.
Read the case studyThe Unicorn Advisory left Teachable after sales pages and support stalled
She copied Business 101 onto Kajabi by hand. Teachable hosted the course. Kajabi won on pages, live chat, and later email and the site. This is a first-person affiliate post from 2019.
Read the case study4ocean left BigCommerce and went back to Shopify
Alex Schulze moved 4ocean back to Shopify after a BigCommerce migration cost organic traffic and turned everyday merchandising into custom work. The public write-up is a Shopify case study.
Read the case studyWorkleap moved acquired Barley from Google Workspace to Microsoft 365
ShareGate's parent used its own tool to move Barley off Google after the acquisition. Files took two hours and mail about three, after days of mapping. This is a vendor case study of the vendor's own product.
Read the case studyBlue Vanilla left WooCommerce maintenance for Shopify Plus
Lucy Colgrave moved a London fashion store off a slow WordPress and WooCommerce stack. Swanky rebuilt it on Shopify Plus. Page load and bounce improved in the first weeks. This is an agency case study from 2019.
Read the case studyCommunity Sauna rebuilt support while leaving Zendesk for Intercom
SupportYourApp moved Community Sauna Baths off Zendesk onto Intercom and rebuilt inboxes, automations, and a chatbot. The 97% faster-response claim is the vendor's. This is an implementer case study.
Read the case studyCompanio moved 56 people to Zoho because Google hit €1,000 a month
Ignacio Nieto Carvajal left Google Workspace for Zoho Workplace to cut a pay-per-user bill. He says Zoho is worse and still worth it. The post is about cost, not the mailbox copy.
Read the case studyAn eight-user company ran Google and Zoho in parallel for ten days
Mo Alhabal moved Digital Era Hub from Google Workspace to Zoho Workplace in about ten days. Shared Drive permissions dropped. Calendar free/busy surprised him. YouTube and Business Profile needed a week-long wait.
Read the case studyA healthcare SaaS moved 906,000 Mixpanel events and left PHI behind
ProductQuant audited four years of Mixpanel data for an unnamed healthcare forms platform, excluded events with patient identifiers, and used PostHog's managed import. This is a consultant case study.
Read the case studyLemon Press moved bilingual financial newsletters off Mailchimp
A 16-person Mongolia newsroom left Mailchimp for Beehiiv after send volume pushed them into extra costs. The public write-up is a vendor case study. It says little about the actual import.
Read the case studyNational Mattress copied a dead Mailchimp list into Klaviyo and opened at 3.2%
Northern moved National Mattress from Mailchimp to Klaviyo after a two-year sending gap. The list came over. The sending reputation did not. This is an agency case study.
Read the case studyVIP Supply Co. left Shopify for B2B company accounts and price lists
Jay Earley moved a restaurant-supply catalog off Shopify because one company could not have many buyers. Access Solutions rebuilt the store on BigCommerce B2B Edition in four months. This is a vendor case study.
Read the case studyCloudSpot moved 700TB and 120 million images off S3
Gavin Wade copied CloudSpot's galleries from Amazon S3 to Backblaze B2 with Flexify.IO. The public write-up is a vendor case study. New uploads stayed on S3 until a one-day catch-up.
Read the case studyNodecraft moved 23TB off S3 in seven hours
James Ross copied game save-states from Amazon S3 to Backblaze B2 with Python, shell, and the two APIs. The public write-up is a vendor case study. Cloudflare stayed in front, which is where the egress saving came from.
Read the case studyA QuickBooks to Xero review after a full year
2022 finished on QuickBooks Online. Xero went live on 1 January 2023, and QBO stayed for a month. Bank rules cut three-account reconciliation from hours to about five minutes.
Read the case studyNetlify plugins do not travel with the repo
He connected the Eleventy repo to Cloudflare Pages in minutes, then spent hours rebuilding Critical CSS, Plausible snippets, headers, and redirects that lived in the Netlify dashboard.
Read the case studyLeaving Netlify because GitHub-user pricing did not fit
A working Astro site left Netlify over per-GitHub-user pricing, then left Vercel four months later when per-seat cost showed up there too.
Read the case studyPagecord moved Rails Active Storage from S3 to Cloudflare R2
Olly Headey copied Pagecord off S3 with Super Slurper, turned on incremental fetch for new objects, and pointed Active Storage at an R2 endpoint plus storage.pagecord.com.
Read the case studyA Rails API hosting bill cut from about $20 to $6
He streamed JawsDB MySQL into Railway in 20 to 30 minutes, pointed Rails at MYSQL_URL, and treated the 70 percent cut as a side-project number, not a production forecast.
Read the case studyA Hugo blog moved to Cloudflare Pages for Japan latency
He followed Cloudflare's Netlify guide, pinned HUGO_VERSION, and swapped Netlify's $URL for $CF_PAGES_URL. The speed he felt came mostly from caching fonts, not from the host swap.
Read the case studySaarza left MongoDB when relationships started leaking
Saarza's own blog says schemaless MongoDB left orphaned invoices on their SaaS platforms. They moved to PostgreSQL and Drizzle so foreign keys could reject a delete that still had children.
Read the case studyTraffic jumped after leaving Google Analytics for Plausible
He dual-ran Plausible next to Google Analytics on freshman.tech. Page views came in 30 to 40 percent higher, which he blamed on ad blockers hitting the GA script.
Read the case studyA Swift Vapor app left Heroku, and Redis stayed behind
After Heroku killed the free tier, a Vapor app moved to Fly.io, fly.toml was rebuilt, and the Heroku Redis cache was replaced only later.
Read the case studyNumeric Citizen replaced Google Analytics after a 30-day Plausible trial
JF Martin left Google Analytics on two blogs after a 30-day Plausible trial, paid $4 a month, and cancelled the Databox dashboards that sat on top of GA.
Read the case studyAxiom flipped its public community from Slack to Discord
Alan Pope gave users a month's notice, then shut Slack. Discord's member count passed Slack in two months. This is Axiom's own post, not a vendor page.
Read the case studyCorsearch spent most of its Microsoft-to-Slack budget on change management
John Mazotis moved Corsearch off Microsoft onto Google Workspace and Slack in three months. Copying files was easy. Teaching people to leave channels was the hard part.
Read the case studyEuna Solutions moved Slack to Teams in two weeks
BITSUMMIT says it moved 500-plus people, 200-plus channels, and about 1 TB off Slack in two weeks. This is a vendor case study. Microsoft's own tool still skips DMs.
Read the case studyLeaving ActiveCampaign for Kit after the price jumped
She left ActiveCampaign Lite after a $28 monthly hike, used ConvertKit's free migration team, and said the switch saved $56 a month versus the new ActiveCampaign bill.
Read the case study18 Medium posts moved to self-hosted Ghost
He pasted Medium's HTML export into Ghost on AWS Lightsail, rehosted the images, and flipped DNS after testing one Cloudflare bulk redirect.
Read the case studyLeaving Ghost for Astro in about a day
He exported Ghost, converted posts to Markdown, rebuilt the site in Astro and Tailwind, and shipped it to Cloudflare Pages. Redirects took more time than the export.
Read the case studyA new Ghost site, with the WordPress blog kept as an archive
He tried a full WordPress import, then left the old site live. New writing lives on Ghost Pro as Studio6 at nevillehobson.io.
Read the case studyMirroring 15 years of WordPress before converting it to Astro
He kept a wget copy of the live site, exported Markdown with Jekyll Exporter, and rebuilt the blog as static Astro on Azure's free tier. The exporter needed cleanup, and plugin features had to be rewritten.
Read the case studyA 10-year WordPress blog moved into local Markdown and Astro
A server move felt like too much work for a mostly static site. He converted hundreds of posts and pages to repo Markdown, kept the old URLs, and cached HTML at Cloudflare.
Read the case studyOscar Health left one of the world's most complex Jira instances
Pete Zalewski moved 600+ people from a maxed-out Jira instance to Linear in just over a month. The public write-up is a vendor case study. They imported history and left custom fields behind.
Read the case studyReadMe left Heroku for Render with about 90 seconds of downtime
Ryan Park moved an eight-year Heroku estate onto Render. The public write-up is a vendor interview. Hard downtime was the MongoDB primary promotion in Oregon.
Read the case studyFingers moved from Substack to Beehiiv in 90 minutes
Dave Infante took the paid list and most of a four-year archive. He left comments, some galleries, and more than 2,000 silent free addresses behind.
Read the case studyLeaving Substack for Ghost and paying more than expected
They moved Amplify Respect onto Ghost Pro after a self-hosted detour, kept the custom domain, and left comments on Substack because those do not come across.
Read the case studyCurating Evernote into Obsidian instead of dumping 12,000 notes
He copied frequent notes and keepers by hand, moved the rest only when he needed them, and kept Evernote as a fallback until a later archive export.
Read the case studySpring Framework moved 15 years of Jira issues to GitHub
Rossen Stoyanchev imported 17,000 Jira issues into GitHub. Markup conversion hurt, and the import API emailed millions of people before GitHub support cut notifications off.
Read the case studyUnwindHR moved from Firebase to Supabase
Joe Ashwell took UnwindHR offline for a weekend, rebuilt the data model in Postgres, and left Firebase after nested reads and lock-in started to hurt.
Read the case study