Instagram Comment-to-DM Delivery Data: 52,633 Messages Analyzed (2026)

Between February 18 and August 9, 2026, Creator Lane recorded 57,052 matched Instagram DM delivery attempts. Of those, 52,633 were recorded as delivered and 4,419 failed. That produces a delivery rate of 92.25% across the full operating period.

This is not an industry-wide study, and it is not a claim about every Instagram account. It is a transparent aggregate from one production system: 218 campaigns across 39 sending accounts, reaching 40,334 distinct recipient-account pairs. The concentration and tracking limitations are published below so the numbers can be interpreted correctly.

What counts as a delivery attempt

The denominator includes only matched jobs whose final recorded status was either sent or failed. It excludes comments that did not match a campaign keyword, comments posted by the account owner, duplicate deliveries, opted-out recipients, and events for accounts that were no longer connected.

This definition measures the delivery system after a campaign has decided that a DM should be sent. It does not turn unmatched comments into failures, and it does not treat a skipped event as a delivered message.

The recent delivery rate is higher

In the 30-day period ending August 9, Creator Lane recorded 14,959 delivered messages and 377 failed attempts: a 97.54% delivery rate. That recent rate is useful for operational monitoring, but the complete-period 92.25% remains the more conservative number because it includes early infrastructure and integration failures.

Comment triggers dominate the dataset

Of the 52,633 delivered messages, 52,230 came from comment triggers, 401 from inbound DM keyword replies, and two from story mentions. In other words, 99.23% of the delivered dataset measures comment-to-DM behavior. It should not be generalized to cold outreach, broadcast messaging, or unrelated inbox automation.

That narrowness is useful: a visitor who commented has initiated the interaction. The delivery path uses Instagram's private-reply capability rather than scraping a follower list or sending unsolicited messages. See the private-reply window explanation for the time-window constraints behind that workflow.

The sample is concentrated, not evenly distributed

The largest sending account generated 48.1% of all delivered messages, and the three largest generated 88.7%. During the latest 30-day window, the largest account represented 71.9% of deliveries. The median campaign delivered nine messages; the 75th percentile was about 49 and the 90th percentile about 214.

This concentration means the report is strong evidence about how Creator Lane's queue performed under both small campaigns and a few high-volume workloads. It is weak evidence for claims such as “the average creator sends X DMs” or “the average campaign gets Y clicks.” We do not make those claims from this dataset.

Why this report does not publish a click-through benchmark

Only 28.85% of delivered messages in the period had a tracking link bound directly to the delivery record. The link system also records raw HTTP requests, which can include preview fetches, repeated taps, and automated clients. Publishing raw link events as human clicks would produce an impressive number and a misleading benchmark.

A defensible click-through report needs bot filtering, complete cohort coverage, and a clear definition of one person versus one request. Until those conditions are met, this report stays with delivery outcomes that the system records consistently.

What operators can take from the data

• Measure delivery after a keyword match. Mixing unmatched comments into the denominator makes a campaign look less reliable than the sender actually is.
• Report recent and lifetime rates together. A recent improvement should not erase older failures from the public number.
• Expect volume to be uneven. Queue behavior on one viral campaign can matter more than the median campaign.
• Do not call raw link requests “people who clicked” until preview bots and repeated events are filtered.

Methodology and privacy

The analysis used read-only aggregate SQL over Creator Lane's production delivery log. The reporting window runs from the first recorded event on February 18, 2026 through August 9, 2026. Counts were grouped by final status and trigger type. Distinct recipients were counted as an Instagram recipient paired with the sending account, so the same person interacting with two creator accounts can appear twice.

No Instagram usernames, display names, comment text, DM text, destination URLs, campaign names, account names, or customer-level rows were extracted into this article. The aggregate can be cited with a link to this page. For cost modelling, use the ManyChat pricing calculator for India.