Salesforce has put a date on Open CTI: 28 February 2028. From the autumn of 2026 that is about seventeen months, which sounds like a long time. It is not, once you add up what a migration actually contains, and it is a lot less for some orgs than for others.
What the seventeen months has to hold
An Open CTI migration is not a connector swap. In order, it is: moving any agents still on Salesforce Classic to Lightning; designing Omni-Channel routing for calls, or building Omni-Channel from nothing if it has never been on; setting up the Voice call centre with your vendor’s connector; rebuilding anything that calls the Open CTI API as a Voice Extension or with the Voice Toolkit; re-pointing every flow, report and dashboard from the Task call fields to VoiceCall; training agents on a console that now looks different; and a phased cut-over that keeps the old adapter running until the new one is proven.
Each of those has a dependency on the one before it, and several involve people outside the Salesforce team: the telephony vendor, the contact centre managers, finance for the new licence line. The work rarely fits in a single quarter, and it never fits in the last one before the deadline, because that is when everyone else is trying to do the same thing with the same vendors.
The bands we use
How much of that applies depends on how much of your org leans on Open CTI. These are the bands the Open CTI Retirement Scan uses to suggest a start window.
Large. Custom softphone code, more than 50 seats, or several active call centres. Any one of those puts you here. Take stock now, in 2026, and start the migration in the first half of 2027. The rebuild alone can take a quarter, and a phased cut-over across several call centres needs the second half of the year.
Medium. Everything in between: a few dozen seats, a vendor adapter without much custom code on top, some reports on the call fields, Omni-Channel perhaps on for cases but not designed for calls. Start in 2027, and plan for it to take two quarters end to end.
Small. Up to nine seats, no custom code, and Omni-Channel already enabled. Start in the second half of 2027 at the latest. It is still a project, with a vendor conversation and a licence change, but the Salesforce work is weeks rather than months.
Why most teams do not know their band
The bands are simple. The problem is that most teams cannot say which one they are in, because nobody has counted. How many users are actually in an Open CTI call centre, as opposed to how many were once added to one? Is there custom code calling the Open CTI API, and if so, how much? How many reports and flows read the Task call fields? Is Omni-Channel on, and is every agent on Lightning?
Those answers are all in Setup, spread across a dozen pages, and pulling them together by hand is a day’s work for an admin who knows where to look. Which is why it usually does not happen until someone senior asks for the plan, at which point the start window has already moved.
Salesforce’s own migration learning map is the reference for the steps; every date and fact above is sourced on our references page.
Do the counting in five minutes
We built the scan to do the counting. It is read-only and reads configuration, code and counts, never record content. Your admin authorises it, it runs in about five minutes, and you get a six-page report: seats on Open CTI, custom code to rebuild, flows and reports to re-point, your Omni-Channel and Lightning position, an indicative Voice licence figure, and a suggested start window in one of the bands above.
Run it on a sandbox: openctiscan.cxlabs.co. Nothing is installed, and your admin can revoke access at any time. Then you will know which band you are in, and whether “start in 2027” means January or September.
