Data report · Q3 2026 · published 14 September

State of Instagram DM booking, Q3 2026

This is what happened in every Instagram DM conversation Setterr handled for its live workspace — a sports-rehab clinic — between 1 August and 14 September 2026, aggregated and anonymised. 499 engaged conversations; 29.5% reached qualified; 11.4% booked a call; the median AI first reply took 52 seconds. The numbers, the cuts, the caveats, and the CSV.

Headline figures

For reference, the only published industry average we know of — SetSmart's 2026 study of 828,000 conversations — puts engaged-to-qualified at 22.9% and engaged-to-booked at 1.94%. Their definitions may differ from ours; see methodology.

By month

July, the workspace's first month, is excluded: the owner was still editing the playbook and the AI's brief, and the booked rate that month (9.1% of 372) reflects tuning rather than the settled setup. September is a half month with some conversations still open.

By source: click-to-DM ads versus organic

Ad conversations booked at a higher rate than organic ones once both were answered in seconds — the opposite of the usual pattern, where ad leads arrive in bursts, get answered late, and underperform. With the delay removed, the ad's ice-breaker question does its job as a first qualifier. Attribution comes from the ad referral Meta attaches to click-to-DM conversations; see the click-to-DM guide.

By first-reply time

An honest limit of this dataset: 94% of first replies were under two minutes, so the slower buckets are too small to draw the speed curve from — six conversations in total waited more than ten minutes. What the data does show is the baseline: when nearly every conversation is answered in about a minute, roughly one in ten books. The research on what happens as replies slow is summarised in DM response time; we'll add our own curve when we have enough slow replies to measure one, which we hope never to have.

Follow-ups: the finding we didn't expect

Read this carefully, because it's easy to misread. Follow-ups go only to leads who went quiet, so the "none" row is the leads who stayed in the conversation, and their 26.6% is not a follow-up effect. The number that matters is the last row: of 277 leads who went quiet and received two or more automated follow-ups, none booked. Sales lore says most calls come from the second touch onward. In this dataset, automated DM follow-ups recovered no one; every booking came from a conversation that never went quiet.

Our reading: on Instagram, a lead who stops replying mid-conversation has usually decided, and a nudge doesn't change that. Follow-ups cost nothing to send, so there's no reason to stop, but the leverage is entirely in the first reply and the qualification, not the chase. We've updated our own guides to say so. Whether a different cadence or copy would do better is an open question we'll test.

Thread length, and what "1,000 messages" buys

The median engaged conversation ran 14 messages; the median booked conversation ran 58. Across 499 conversations there were 11,895 messages, 24 each on average. Tools that price per message (a 1,000-message allowance, say) therefore cover roughly 42 conversations at this mean, or under 20 conversations that actually book. Tools that price per conversation cover the same number regardless of thread length. Worth knowing when you compare plans.

What we can't tell you yet

• Show rate. Booking statuses weren't maintained by the tenants this period (149 of 152 bookings still read "confirmed"), so we can't report how many calls happened. We're adding a lighter way to record it.
• Time of day in local time. We can say 25% of inbound messages arrived on a Saturday or Sunday; we won't guess local hours from UTC.
• The speed curve. As above: nearly everything was fast, so we can't measure slow.
• Generalisation. One workspace, one niche (a sports-rehab clinic), six weeks. This is a report on Setterr's live workspace, not the industry. Further workspaces are added as they pass their onboarding period.

Methodology

• Scope: every conversation created between 1 August and 14 September 2026 in Setterr's production database for the workspace that has been live since July. July itself is excluded as the tuning month (the owner was still editing the playbook). Workspaces still in their onboarding period, internal test accounts, the end-to-end test user, and "test drive" threads (where an owner messages their own setter) are excluded.
• Engaged: at least one inbound message from the lead and at least one outbound reply (AI or human).
• Qualified: the conversation reached the "qualified" or "booked" stage in the owner's playbook, or has an attributed booking. Stages are set by the AI when the owner's qualifying questions are answered and criteria met; owners can override.
• Booked: a non-cancelled booking record attributed to the conversation (Calendly sync or owner-confirmed), or the conversation stage is "booked".
• First reply: seconds from the lead's first message to the first outbound message after it, by whoever sent it.
• Follow-ups: the count of automated follow-up messages the conversation received, recorded per conversation.
• Anonymisation: aggregates only. No names, handles, message text or per-lead rows leave the database.
• Revisions: figures are computed at publication and will be revised in the next quarterly report, with changes noted here.

Download and cite

All aggregates above: instagram-dm-booking-2026-q3.csv. Licensed CC BY 4.0 — quote freely with a link. Suggested citation: Setterr, "State of Instagram DM Booking, Q3 2026", setterr.com/state-of-instagram-dm-booking-2026/, 14 September 2026. The benchmarks page keeps the headline figures current and has a calculator.

Frequently asked