Data we see across SaaS accounts is consistent: the probability that a customer is "support-active" at 12 months is better predicted by their day-7 behavior than by your product quality. Uncomfortable, but actionable.
Day 0: what should the welcome message contain?
A message that says "welcome, feel free to reach out" is useless. What works: a message that anticipates the most likely first question based on the product step. "You just created your workspace. The next step is X, here's how, and if you want to talk about it, reply here."
Day 1 to 3: how do you get the first interaction right?
- Goal: get a reply on a real question, not a test ping.
- Lever: detect the first session without a completed action, trigger a contextual message.
- Guardrail: max 1 message per customer in the first 72h. Beyond that it's harassment.
Day 4 to 7: what goes in the check-in?
An effective check-in contains:
- A personalised reference to what the customer has (or hasn't) done.
- An open question, not a 5-star survey.
- An easy out if they don't want to reply.
Target reply rate on this check-in: 25 to 35%. Below that, the format is too generic. Above, you're probably hitting a measurement bias.
Which anti-patterns kill support onboarding?
- Sending everything via automation with no named person in signature. The relationship never installs.
- Asking for NPS on day 3. You're measuring mood, not satisfaction.
- Handing onboarding to FIN with no human backup. On first interactions, AI alone scores negatively even when the answer is correct.
Which metrics should you track over the first seven days?
- Time to first reply by customer (TTFR-c), days before the customer reaches out unprompted.
- Day-7 activation rate on support-engaged vs non-engaged cohort.
- Average CSAT on day 0-7 conversations vs the rest of the journey.
Support onboarding is not marketing onboarding with a nicer tone. It's the highest-leverage investment for 12-month retention.