Podcast: Download (Duration: 50:57 — 58.6MB)
Using Instagram DMs for ecommerce works by replacing your swipe-up call to action with a DM keyword prompt, which in Natasha Takahashi’s testing generated 500% to 800% more leads than sending the same audience to a link. The mechanism is micro-commitment: each step someone takes inside the conversation increases the chance they complete the purchase.
In this episode I brought Natasha Takahashi back on the show. She is founder and CMO of School of Bots and has been testing Instagram DM automation with clients since early 2021.
Below is the practical layer: how the campaigns are actually built, the product recommendation quiz, capturing emails inside chat, how DM differs from Messenger, and how to handle handoff to a human.
Get My Free Mini Course On How To Start A Successful Ecommerce Store
If you are interested in starting an ecommerce business, I put together a comprehensive package of resources that will help you launch your own online store from complete scratch. Be sure to grab it before you leave!
Table of Contents
Key takeaways
- DM prompts beat swipe-up links by 500% to 800% on leads generated, despite adding friction.
- 50% of DM conversations with businesses start from stories.
- 81% of Instagram users search for products and services on the platform.
- Three-story sequences get the most engagement and prime people for the offer.
- Asking for an email optionally captures 50% to 80% without deterring the rest.
- A product recommendation quiz is the highest-value asset you can build for ecommerce.
- Keep keywords under seven characters so people actually type them correctly.
- Instagram has no sponsored messages, so capture email or phone during the conversation.
Why a DM prompt outperforms a swipe-up link
Telling people to DM you a keyword generates 500% to 800% more leads than sending the same audience to a link, which held across every test Natasha ran.
This is counterintuitive because typing a word is more friction than tapping a link.
The explanation is where conversations already happen. Half of all DM conversations between people and businesses on Instagram start through stories, so the reply bar is a behavior people already have.
The platform context supports it. Instagram reports that 81% of regular users search for products and services there, which makes it a buying environment rather than only a social one.
The friction turns out to work in your favor. Each step someone takes inside a conversation is a micro-commitment, and people who commit progressively convert at higher rates than people who tap through to a page they can close instantly.
That last point matters specifically on Instagram. A swipe-up opens a web view inside the app that is trivially easy to dismiss, and once someone exits you have no way to follow up beyond retargeting.
How to structure an Instagram story campaign
Build a sequence of at least three stories that surface the customer’s problem, then close with a DM keyword prompt on the last one.
Instagram’s own data shows sequences of three or more stories get the most engagement, and the sequence also puts the viewer in the right frame of mind for the offer.
Natasha’s Mother’s Day example: three stories asking about the viewer’s situation or pain point, structured as polls or this-or-that questions, leading naturally to a conclusion.
The fourth story closes it. Something like acknowledging they still need a gift, announcing 15% off, and telling them to DM the word “mom15” for a code good that day.
Keep the keyword short. Under seven letters or numbers is the rule, because people have to type it correctly for the automation to fire.
How to capture emails inside the conversation
Ask for the email optionally rather than requiring it, which captures 50% to 80% of leads without losing the people who only want the discount.
The phrasing matters. Offer to send the coupon to their email as well so they have it, and make clear that if they would rather not, they can move on and still receive the code.
That optional barrier is deliberate. Requiring an email to get a discount deters people who came for the discount, and asking for nothing captures nobody.
For a product under about $50, this is enough. Elaborate nurturing is unnecessary when the purchase decision is small.
How to build a product recommendation quiz
A product recommendation quiz is the single most reusable asset for an ecommerce DM strategy, because it slots into the front or back of any campaign.
The structure is multiple choice throughout, never open-ended. You ask about the customer’s pain points, challenges, or goals, then recommend a specific product at the end.
The threshold for it making sense is roughly five to ten products, or at minimum more than one or two. With a single product the quiz becomes thin and reads as a formality.
Personalized product categories already do this well. Curology building a custom skincare formula is the model, and many ecommerce brands now run the same quiz on their websites, which makes porting it into DMs straightforward.
True-or-false quizzes convert worse. They work for engagement and for learning what your audience wants, and giving someone an actual recommendation is what drives the sale.
The cadence Natasha recommends: run a story once a week driving new traffic into the same quiz, then attach a cart abandonment sequence behind it. That combination is one of the fastest ways to see ROI from DM automation.
How to follow up after someone clicks
Message again 10 to 30 minutes after someone clicks your link, while you are still inside the 24-hour window.
The follow-up is simple: ask whether they found something they liked, or whether they had a chance to check out.
If you captured the email, cross-reference it against your store data so you do not send a purchase nudge to someone who already bought. Thank those people instead and route them somewhere useful.
Anyone who says they have not bought yet is a natural candidate for the product recommendation quiz.
How Instagram DM differs from Messenger
Four differences matter, and most of them simplify rather than restrict.
Message formats. Instagram supports quick reply buttons, open user input, images, and GIFs. Messenger additionally offers card galleries, richer media, and integrations. Natasha considers the reduced set a benefit, since Messenger’s option count left many businesses unsure where to start.
Opt-in methods. Messenger has a unique link that opens a conversation directly. Instagram has no equivalent, so people must reach your profile and message from there. Ads that prompt a DM are the exception.
Follow-up options. Messenger has message tags and one-time notifications for messaging beyond 24 hours. Instagram has only the 24-hour window.
Sponsored messages. Messenger has them and Instagram does not.
For EU accounts specifically, the only Instagram restriction is sending media. Images, GIFs, and video are unavailable, and everything else works.
Why Instagram beats Messenger for organic traffic
The conversion mechanics are comparable and the traffic volume is not, which is where Instagram wins.
Natasha’s tests found similar click rates, retention, and conversion when the same campaign ran on both platforms.
The difference is re-engagement. Stories are watched constantly, arguably more than the feed, which means you can pull people back into a conversation repeatedly without paying for it.
Facebook stories often get more viewership and Instagram stories get more engagement, which is the distinction that matters for starting conversations.
The organic reach advantage is partly a numbers game. Instagram is the smaller platform, and smaller platforms deliver more organic reach until they mature.
How to re-engage past the 24-hour window
Without sponsored messages, re-engagement happens through organic content or paid posts that prompt a new DM.
The organic route is stories, feed posts, and reels carrying a fresh keyword prompt.
The paid route works too. Run a feed ad or a story ad where the call to action is to DM you, and anyone who responds re-enters the window.
You can build custom audiences from your subscribers. A tool like ManyChat can add each new subscriber to a Facebook custom audience automatically, which makes targeted re-engagement ads practical.
Natasha’s framing of this is that it is outbound-adjacent rather than a true sponsored message, since you are prompting people to message you instead of messaging them.
Her broader prediction is that the model of free subscribers plus paid messaging will not survive alongside email as more messaging channels open up.
How reels fit in
Reels have no automation trigger, so the workaround is putting the keyword prompt in the caption or on-screen text.
The same applies to IGTV. Only feed post comments and story replies currently trigger automation.
The side benefit is measurement. Because each reel can use its own keyword, you can attribute DM volume to specific reels and gauge what they are actually worth.
How to handle handoff from bot to human
Answer as much as possible automatically, ask whether the answer helped, and route anyone who says no to a live person.
The system Natasha builds has three stages. An automated support layer answers common questions using keywords plus AI, then the bot asks whether that answered the question.
Anyone answering no becomes an open conversation.
Notification routing keeps it manageable. Open conversations trigger a Slack alert, and support staff respond inside ManyChat the way they would in Zendesk or Intercom.
The handoff reverses automatically. Once the issue is resolved or the person has gone quiet for a day or two, the conversation closes and the contact returns to the bot.
Business hours can be built into the logic so the routing only happens when someone is available.
When to add AI to your chatbot
Layer Dialogflow on top of ManyChat rather than replacing it, because each handles a different job.
Keyword matching in ManyChat catches predictable phrasing. Dialogflow learns to recognize that differently worded requests mean the same thing, then hands control back to ManyChat to execute.
The pizza example illustrates it. “I want to order a pizza” matches a keyword. “I’m looking to buy a pizza with pepperoni delivered at 5pm” likely does not, and Dialogflow can learn to recognize it as the same intent and confirm with the user.
The reason not to move everything to Dialogflow is that it is not a marketing platform. It has no CRM and no content management, so ManyChat remains the base holding your subscribers, content, and automation.
To the customer the seam is invisible. They experience one conversation regardless of which system generated a given message.
Where Instagram DM automation is heading
Facebook’s stated goal is interoperability across Messenger, Instagram DM, and WhatsApp, meaning a consistent messaging experience rather than one merged app.
The practical implication is that Messenger is a preview of Instagram’s roadmap. Features that exist there are reasonable predictions for what arrives on Instagram.
At the time of this conversation, Instagram had roughly 10% to 20% of its planned feature set live, with more rolling out as Facebook gathered data.
Contact lists were still separate. The same person with both an Instagram and a Messenger profile appeared as two subscribers, with contact merging expected later.
The area to watch is shopping. Commerce is Instagram’s stated priority, and triggers tied to shop browsing or product taps are the likely next batch, which is where Natasha expects most ecommerce ROI to come from.
Frequently asked questions
How do you use Instagram DMs for ecommerce?
Replace swipe-up calls to action with a DM keyword prompt, then run the offer inside the conversation. Natasha Takahashi’s testing showed 500% to 800% more leads generated this way, because each step in a conversation is a micro-commitment that increases conversion.
Why do DM prompts outperform swipe-up links?
Half of all business DM conversations on Instagram start through stories, so replying is an existing behavior. A swipe-up opens an in-app web view that is easy to dismiss with no follow-up path, while a conversation lets you message again within 24 hours.
How many stories should a campaign use?
At least three before the call to action. Instagram’s data shows sequences of three or more get the most engagement, and using them to surface the customer’s problem primes people for the offer that follows on the final story.
Should you require an email to get a discount code?
No. Ask optionally and make clear people can decline and still receive the code. That approach captures 50% to 80% of emails without deterring people who only want the discount.
What kind of quiz works for ecommerce?
A multiple choice product recommendation quiz that asks about pain points and goals, then recommends a specific product. It needs roughly five to ten products to make sense. True-or-false quizzes work for engagement and convert worse than actual recommendations.
Can you send broadcasts on Instagram DM?
No. Instagram has no sponsored messages and no message tags, so you can only message within 24 hours of someone’s last interaction. Capture an email or phone number during the conversation, and re-engage through stories, posts, or paid ads prompting a new DM.
Do you need AI for an Instagram chatbot?
Not to start. Keyword matching handles predictable phrasing. Adding Dialogflow on top helps recognize differently worded requests with the same intent, while ManyChat remains the base holding subscribers, content, and automation.
