@aftershower launched a product on Instagram on 18 August 2026 with one Reel and one instruction: comment a keyword to get the details in your DMs. The campaign was set up and run by the brand's agency on UnlockDM.
By the end of launch day, 702 people had commented to enter. The campaign collected 750 entries in total and sent 936 automatic DMs before it was paused four days later.
This is real campaign data, shared with permission. It is one launch, so treat it as a worked example rather than a benchmark. It is useful because it behaves so differently from an evergreen campaign.
The campaign in numbers
| Metric | Value |
|---|---|
| Entries on launch day | 702 |
| Entries in total | 750 |
| DMs sent (private replies and rewards) | 936 |
| Public comment replies | 740 |
| Messages that failed to deliver | 30 (under 2%) |
| Days live before it was paused | 4 |
The keyword was the Hebrew word for shower, which fits the product and is exactly what the audience would type.
A launch spikes, then stops
Almost everything happened in the first 24 hours.
| Day | Entries | Share of total |
|---|---|---|
| 18 Aug (launch) | 702 | 94% |
| 19 Aug | 39 | 5% |
| 20 to 22 Aug | 9 | 1% |
Compare that with @bhipakshiii's AI-tools Reel, which got only 31% of its entries on day one and was still collecting 60 or more a day in its fifth week.
Inference from the two campaigns: the difference is the content, not the tool. A tutorial or resource Reel stays useful, so people keep finding it through search and shares and keep commenting. A launch announcement is news. Once the launch has passed, there is little reason to come back to it.
That changes how you plan:
- Evergreen Reel: expect a long tail, leave the campaign running, and size it for weeks of entries.
- Launch Reel: expect nearly everything on day one, make sure the campaign can handle the spike, and plan what happens when the launch ends.
Lesson 1: rehearse before the real post
The same morning, before the launch Reel went out, the team ran two small practice campaigns on other posts. Between them they collected five test entries: enough to see the automatic reply arrive, check the DM text, and tap the link.
On a launch, the first hour is the one that matters most. Finding a typo in the reward message after 300 people have already received it is too late. A ten-minute rehearsal on a low-stakes post is the cheapest insurance you can buy. Check three things: that the reply arrives, that the wording reads well on a phone, and that the link opens the right page.
Lesson 2: use the word your audience would type
It is tempting to pick a clever or branded keyword. The better choice is the word people would say anyway. Here, that was a single everyday Hebrew word tied directly to the product, and it worked: 750 people typed it.
Two practical points:
- Any language works. Comment keywords are matched as text, so Hebrew, Hindi, Indonesian, or Spanish work the same way English does.
- One word beats a phrase. The fewer characters someone has to type on a phone, the more of them will do it. Our guide to keyword calls to action goes deeper on this.
Lesson 3: plan the end of the launch
This campaign was paused four days after launch, once entries had dropped to almost nothing. That is the right move for a time-bound offer. A launch discount or early-access link that keeps going out after the launch ends turns a good first impression into a confused one.
If the post might keep getting views, the alternative is to update the reward to whatever is still available instead of pausing. Our guide to running a waitlist or product launch through DMs covers building a destination page that still works after launch day.
Why automation matters most on a launch
702 people in one day is roughly one comment every two minutes, around the clock. On a launch day, the team is also answering questions, watching sales, and posting stories. Replying to each comment by hand would have been impossible, and every late reply would have lost some of the launch-day excitement.
With the campaign running, those comments got a private reply automatically; only 30 of more than 1,700 messages failed to deliver. The team's only job was the launch itself, and the people who commented at two in the morning got the same fast answer as the people who commented at noon.
More campaign results
- One Reel, 3,419 entries: an AI-tools creator whose campaign was still collecting entries five weeks later.
- One campaign per product: a Shopee affiliate creator, and why a product-name keyword beat a generic one.
Sources checked
Checked on 2026-09-25.
- UnlockDM production data for @aftershower's launch campaign, 18 to 22 August 2026, shared with permission
- Comparison figures from @bhipakshiii's campaign, 24 August to 25 September 2026
UnlockDM runs comment-to-DM campaigns for launches and evergreen posts alike, in any language, priced per campaign instead of per month. For a launch, rehearse on a practice post first and decide in advance when the campaign ends.
