How Finnish studio Kovaa moved 180 sites to Rivervo in one weekend.

Kovaa is a digital studio in Tampere with about 180 client sites — mostly WordPress, some static, a handful of Laravel. Over the weekend of February 7th–8th, they moved every single one off their previous host and onto Rivervo. We sat down with Janne Korhonen, Kovaa's lead ops engineer, to talk about how it went.

This interview has been edited for length and clarity.

"We gave ourselves 52 hours."

Rivervo: Why move all 180 sites in one weekend instead of staging it over a month?

Janne: Honestly? I've done staged migrations before and they're miserable. You end up with two billing relationships, two sets of credentials to manage, half your monitoring pointed at the old host and half at the new one. For a month you're the worst version of your operations team. If it's going to hurt, I'd rather it hurts concentrated.

Also — and this matters — our previous host had a renewal on the 10th. Staging would have meant paying for both services. A clean cutover before the renewal saved us €14,000.

"The first thing we did was nothing."

Rivervo: Walk me through the weekend. When did work start?

Janne: Friday at 16:00 we sent the email to clients telling them DNS propagation might cause brief issues over the weekend. Then we did nothing until Saturday morning — deliberately. I wanted the DNS TTL reduction we'd set 72 hours earlier to have fully propagated.

Saturday 08:00 we started the first batch. Your team had already provisioned everything — all 180 accounts were ready on your side, plans pre-configured, cPanel credentials in a spreadsheet. We just had to run the sync.

"Your migration tool did the heavy lifting."

Rivervo: You used our free migration service for most of them?

Janne: For 147 of the 180, yes. Your tool handled cPanel-to-cPanel copies cleanly. Files, databases, email accounts, SSL certs, cron jobs — all of it. We'd click "Migrate this site," point it at the source credentials, and come back to a working copy.

The other 33 were the non-cPanel ones — the Laravel apps and two sites running on weird custom stacks. Those we did manually. Each one took maybe 30–45 minutes; the Laravel ones were easiest because they were already dockerised.

"The hard part was email."

Rivervo: What went wrong?

Janne: Email. IMAP transfers are slow — you can't parallelise past a certain point without getting throttled, and some of the client accounts had 20 GB of mail going back 15 years. We had to queue those and let them run overnight.

The other thing was a handful of plugins that had the old host's server hostname baked in. One WooCommerce store had a custom reports plugin that made an API call to https://oldhost-reports.example.com/api/... on every checkout. Broke quietly on Saturday. The client noticed Monday morning when reports stopped generating. That was embarrassing.

"By Sunday afternoon it was done."

Rivervo: When did you know you were going to make it?

Janne: Saturday about 22:00. We'd migrated 110 sites by then and the success rate was around 97%. The remaining 70 were mostly smaller sites that would move fast. At that point I went to sleep for six hours and came back Sunday morning to finish.

Final site completed cutover at 14:30 Sunday. We spent the rest of the day monitoring, running automated checks, and catching the two or three sites that needed DNS adjustments.

"What would you do differently?"

Rivervo: If you were starting over, what would you change?

Janne:

1. Run a search for hardcoded hostnames across every site beforehand. grep -r across all client codebases for the old host's domain. That would have caught the WooCommerce plugin issue before it became a Monday-morning problem.

2. Pre-stage the email migrations a week early. Not the cutover — just the initial IMAP copy. Then over the weekend you only need to sync the delta, which is seconds instead of hours.

3. Lower TTLs 7 days out instead of 3. Three days is enough in theory. In practice, some ISPs honor TTLs loosely, and a handful of DNS resolvers held onto the old records for another 24 hours. Not Rivervo's fault — just DNS being DNS.

"The numbers were better than expected."

Rivervo: What's the performance difference been?

Janne: Measurable and real. Across the portfolio, median TTFB dropped from about 380 ms to 165 ms. Our biggest client — a WooCommerce store doing about 2,000 orders a day — saw checkout completion times drop by 1.8 seconds. They noticed within 48 hours.

Support tickets from our clients dropped too, though I can't separate out how much of that is "better hosting" versus "we're not dealing with the old host's support queues anymore."

"Would you recommend the weekend approach?"

Rivervo: For teams considering something similar?

Janne: If you have fewer than 50 sites and you've done two or three migrations before, do a weekend. If you have more than 200 sites or your team has never done this at scale, stage it over 3–4 weekends and pace yourselves.

The thing that made it work for us wasn't anything clever on our side. It was that your migration tool is actually reliable. We didn't have to second-guess whether a site had copied correctly. That's the whole game. Most of the stress in a big migration is "did this work, really?" and your tool made that question answerable fast.

---

Thanks to Janne and the Kovaa team. If you're considering moving sites to Rivervo, we'll do the migration for free — one site or two hundred.