A home and kitchen brand selling into the US market with ten to fifty people on staff sends four hundred creator messages a week. Roughly a third show as opened. About eight creators reply.
The team reads those numbers as a copy problem and rewrites the template. The rewrite changes nothing, because the problem is not the words.
Here is what is actually happening. Four hundred creators receive the same six sentences, none of which mention anything those creators have done. A product card arrives attached to a message that never established why this particular creator should care about a cast-iron skillet. The creator opens it, sees a pitch, and moves on — not because the offer is bad, but because nothing in it was addressed to them.
This article is about the part of bulk outreach that gets skipped: what happens between picking the creators and pressing send. It assumes you already have volume working mechanically and want the reply rate to justify it.
The honest framing up front. Making volume messages sound personal raises reply rates; it does not turn them into individual messages. You are moving from obviously-templated to plausibly-relevant, and that is a real gain, but it is not the same as a human writing four hundred bespoke notes.
“Opened” Is Not the Same as Read
The first thing to correct is the metric.
An open state tells you the message was surfaced to the creator. It does not tell you they read past the first line, and it definitely does not tell you they considered it. Teams that optimise against opens end up writing better first lines, which is a real but small gain.
The number that matters is replies per hundred sent, and the second one is how many of those replies are usable — meaning the creator is interested, not asking to be removed or telling you they do not do home content.
If you are running four hundred a week at eight replies, you are at two per cent. That is not a copy problem at two per cent. At two per cent something structural is wrong, and it is almost always targeting.
Why Volume Messages Get Ignored
Four failure modes, in the order they usually appear.
The message never references the creator. It opens with your brand and your product. The creator has no reason to believe you have seen a single video they made, because you have not said anything that proves it.
The product is wrong for the creator, and volume hides it. Sending home goods to four hundred creators means some meaningful share of them do not make home content at all. They were included because they cleared a filter, not because they fit. Those opens will never convert and they inflate your denominator.
The ask is unclear. The message describes a product and ends. There is no stated next step — no sample offer, no commission figure, no question. The creator is left to work out what you want, and the path of least resistance is nothing.
The second message repeats the first. Follow-ups that restate the original pitch train creators to ignore your name. A follow-up has to carry new information or a different angle, otherwise it is just a reminder that you are still there.
The One Reference That Changes the Response
The single highest-leverage change is one sentence that could only have been written about that creator.
Not “love your content” — that is the sentence every templated message uses, and creators have learned to read it as a tell. Something specific and verifiable: the format they use, the kind of product they feature, a recurring bit, the fact that they review things in a particular way.
This works because it does two jobs at once. It proves a human or a system actually looked, and it gives the creator a reason to believe the fit is real rather than assumed. It also fails loudly, which is useful: if you cannot write that sentence for a creator, they should not be in the send list.
The practical consequence is that your list gets smaller and your rate gets better. Four hundred messages with a real reference each is not achievable by hand. Two hundred, segmented into three groups with three different reference types, is.
The Anatomy of a Message That Gets a Reply
Every message that performs has the same five parts, in this order, and the order matters because the first two decide whether the rest gets read.
| Part | Job | What breaks it |
|---|---|---|
| The reference | Prove you looked | Generic praise, or nothing at all |
| The fit sentence | Connect their content to your product | Describing the product instead of the connection |
| The offer | Commission, sample, or both | Vague “great collaboration opportunity” |
| The ask | One named next step | No ask, or three asks |
| The close | Your name and shop | Brand name only, no person |
Fifty to ninety words is the working length for a first message. Long enough to carry a reference and an offer, short enough to read in the four seconds a creator gives an unknown sender.
Two things teams get wrong inside this structure. They put the offer before the reference, which reads as a pitch with a compliment bolted on. And they make the ask conditional — “let me know if you’d be interested” instead of “want me to send one?” — which converts noticeably worse because it hands the decision back as a formality.
Card Mode and the Quota Argument
There is a mechanical reason to think about message structure that has nothing to do with copy.
Sending a product card and a text message as two separate messages consumes two units of your messaging allowance. Card mode combines the card and the text into a single message, which consumes one. Across four hundred sends a week that is a material difference in how many creators you can actually reach before hitting a ceiling.
The same discipline applies to filtering. Excluding creators you have already messaged, and excluding creators inside a tag you have already worked, prevents the specific failure where a creator gets three versions of the same pitch across two weeks. Repeat contact is the fastest way to get ignored permanently, and it is entirely preventable if the filter runs before the send rather than after.
One caution on around-the-clock sending. Scheduling that runs continuously is useful for reaching creators in different time zones, but a message that lands at 3am local time is a worse message than the same one at 10am. Continuous should mean you can hit any window, not that you use all of them.
What Personalisation Cannot Fix
Three limits, and the third one is the one teams refuse to accept.
It cannot fix a weak offer. A well-targeted message offering 8% on a product with thin margin gets ignored by creators who can see what other brands pay. Segmentation does not compensate for economics.
It cannot fix product mismatch. If your home goods catalogue has four hero SKUs and you are messaging four hundred creators, you will run out of genuine fit long before you run out of creators. The list has to be capped by fit, not by ambition.
And it cannot make a bulk send individually written. Segmented-with-a-reference is meaningfully better than template-plus-name, and creators can still tell. The goal is relevant-at-scale, and pretending it is bespoke is how teams end up overclaiming to creators who then feel misled.
For the mechanics of running volume without tripping platform flags, How Home and Kitchen Sellers Launch Bulk Creator Outreach Without Getting Flagged covers that side, and TikTok Shop Creator Read But No Reply: A Follow-Up Strategy That Gets Responses picks up where this leaves off.
Where DAMI Fits
DAMI sends text, product cards and image messages in bulk, and card mode merges the card and the text into one message — which matters when your monthly allowance, not your ambition, is what caps your reach. Filters run before the send, so creators already messaged, or already inside a tag you have worked, are excluded rather than contacted twice.
The part that connects to the problem above is segmentation. Because you can filter by the same criteria you used to build the list, the message each group receives can carry a reference type that actually applies to it, instead of one paragraph pretending to fit four hundred people.
The limit is the one stated earlier: this improves relevance, not economics. If creators in your category are ignoring you, check what you are offering before rewriting the fourth version of the message.
If you want to test it against your current reply rate, the product page shows how the filters and card mode are configured.
Frequently Asked Questions
How many creators should a first bulk send go to?
Fewer than you think. Cap the list at the number you can write a genuine reference for — for most home brands that is one hundred and fifty to two hundred, not four hundred.
Is card mode always better than a separate card and message?
For first contact, yes, because it halves the allowance cost and keeps the pitch in one readable unit. A separate image message can be worth it later in the conversation when you are showing something specific.
Should follow-ups use the same message again?
No. A follow-up that restates the first message teaches the creator to ignore you. Change the angle, add information, or drop the creator.
What reply rate should we expect?
For a segmented send with a real reference and a clear offer, five to ten per cent is a reasonable target on US home goods. Below three per cent the problem is targeting, not copy.
Does sending around the clock actually help?
Reaching creators in their own evening helps. Sending at 3am because the scheduler can does not. Use the continuous window to hit good times in each zone, not to fill all of them.
