Quarterly industry report based on aggregated, anonymized data from 4,200 message.com workspaces. Five patterns that we did not see coming and we think you should know about going into Q2.
Pattern 1: voice is up 16% YoY
Voice traffic per workspace was up 16% year-over-year, with the steepest growth in healthcare (+34%) and home services (+28%). Chat traffic was flat. Email traffic was down 9%. The shift is real and not just our customers.
Pattern 2: AI deflection rates are converging around 70%
Across our customer base, the median AI deflection rate was 71% in Q1, up from 64% in Q4 2025. The spread is tighter than we expected. The top decile is at 84%, the bottom at 51%, and the spread is mostly explained by KB quality, not industry.
Pattern 3: cross-channel routing is now the default
“85% of conversations now cross at least one channel boundary. Single-channel support is officially dead.”
85% of Q1 chats had at least one cross-channel event (chat-to-email, chat-to-phone, etc.). That number was 62% a year ago. Customers are not picking one channel and staying; they are switching mid-conversation based on convenience.
Pattern 4: response-time pressure is up, not down
Median expected first-response time, based on customer survey free-text, fell from 'within 24 hours' in 2024 to 'within 2 hours' in Q1 2026. AI is teaching customers to expect instant. Even when a human picks up, customers grade harshly above 2 hours.
Pattern 5: enterprise procurement got faster
Average days from first signup to paid Enterprise contract dropped from 91 days to 47 days. We attribute it to two things: SOC 2 Type I in the bag (faster vendor reviews) and customers having mature procurement playbooks for SaaS that did not exist 3 years ago.
What to do in Q2
Three things. (1) Invest in voice. If your team does not have phone integrated, you are leaving 16% YoY growth on the floor. (2) Audit your KB for AI grounding. The top decile is at 84% deflection; getting from 64% to 84% is mostly a KB-quality lift. (3) Set your first-response target at 2 hours, not 24. Customers grade against the new standard, not the old.