Instagram Auto Reply: Story Replies + Comment-to-DM + 24h Window That Recover Sales

I built OT1-Pro because I watched real Instagram sales die in my own DMs while I was asleep or answering the same "price?" comment for the fortieth time that day. In 2025 I ran a unified inbox for Messenger, Instagram and WhatsApp, and the pattern was brutal: stories drove the most ready-to-buy replies, comments created the most missed chances, and anything older than a day went cold. Fixing those three surfaces recovered more revenue than any ad tweak I ever made.

This is my exact instagram auto reply playbook: story replies plus comment-to-DM plus the 24-hour window. I run all three through one inbox at OT1-Pro, with hot/warm/cold scoring, mirrored Arabic and English replies, and human escalation only on buying signals. If you sell through Instagram DMs in Egypt, the Gulf, or anywhere bilingual, this is the setup I use.

If you are still fighting Meta permissions to get the inbox connected, read my winner guide first: Meta App Verification 2026: A Founder's Guide. Without instagram_basic and instagram_manage_messages on Advanced Access, no auto-reply tool can touch your DMs.

Most Instagram auto-reply setups answer everywhere and sell nowhere

I see the same broken setup in store after store. A generic bot replies "Thanks for your message! We will get back soon" to every DM, comment and story reaction. Response time looks good. Sales do not move. I know because I shipped that version first and watched it fail.

The problem is not speed. A story reply asking "Is this still available in medium?" is not the same as a fire-emoji reaction. A comment saying "PRICE" is not the same as one saying "nice". A DM from 20 minutes ago is not the same as one from three days ago. One generic reply cannot handle all three.

When I split my own traffic, story replies closed at 3-4x the rate of cold DMs, comment-to-DM handoffs produced about 60% of new DM threads for fashion shops, and follow-up inside 24 hours recovered 25-35% of stalled threads. My rule now: every auto-reply must catch a story buyer, move a comment into a DM, or pull a warm thread back inside the window. Anything else gets deleted. For the broader picture see Instagram DM inbox for sales teams.

Surface 1: story replies and mentions — the highest-intent trigger

Nobody replies to a story by accident. A feed scroll is passive. A story reply means they stopped, tapped, typed and sent. In my shops that is the closest thing Instagram has to a customer walking up to the counter, so I prioritize it above everything.

I handle three triggers differently. Direct story replies with text ("price?", "بكام؟", "عندكم مقاس لارج؟") get a full context reply naming the item: "That beige set from today's story — yes, medium and large are in stock. Want me to hold one? Send your city for delivery time." Story mentions (a follower tags my shop in an unboxing or try-on) get a short public thanks plus a DM: "Saw you tagged us — thank you! Want 10% off your next order for the share?" Quick reactions with no text get one light line and no catalog dump. Real text earns the full reply. Emoji alone gets a light touch.

The failure I see most: shops blast five paragraphs plus a price list at someone who tapped a heart. They never reply again. Throttle reactions hard and spend your words on buyers who typed words.

Surface 2: comment keyword to DM handoff — where discovery turns into pipeline

Egyptian shops taught me this surface. A post gets 300 comments saying "السعر؟", "بكام", "PRICE", "details", "مكانكم فين؟" and five DMs. Wait for buyers to DM first and you lose 295 chances. Comment-to-DM moves them.

I watch tracked posts and Reels for a short buying-keyword list in both languages: PRICE, details, order, link, بكام, السعر, تفاصيل, عايز, اطلب, توصيل, شحن — plus transliterated "bkam" and "el se3r", because half my buyers type Arabic in Latin letters. A match triggers the handoff: one short public reply ("Sent you the details in DM — check requests") plus a DM within 60 seconds naming the post, price and sizes with one question: "That reel with the black abaya — 1,450 EGP, sizes S to XXL. Which size should I check for you?" Anything without a keyword gets a like or a one-line public reply, nothing more.

I cap it to stay inside Instagram rules: one public reply per comment, one DM per commenter per post, no repeat DMs on silence. Every comment moved to DM becomes a scorable thread I can follow up and close. Public comments rot. DM threads compound. On vendor pricing for those threads, compare OT1-Pro pricing with OT1-Pro vs WATI before you scale this.

Surface 3: the 24-hour window — where sales are actually recovered

On Instagram's Messaging API a business can send free-form messages for roughly 24 hours after the customer's last message. After that you are limited. Almost every lost sale I audited died inside that window from silence, not rejection: buyer asks at 11pm, shop answers at 9am, buyer asks about large, shop answers at 2pm, buyer goes quiet, thread ages out.

I run three timed nudges inside the window. Nudge one at 60-90 minutes: "Holding the large for you — should I pack it with cash on delivery to Cairo?" Nudge two at 6-8 hours: "Two larges left from that story batch. Want me to hold one, or should I release it?" Nudge three at 20-22 hours: "Closing today's hold list — reply YES and I will keep yours until tomorrow." Each stops permanently on any reply, opt-out or human takeover. I never send more than three, never attach links twice, and only mention a discount on nudge three for hot threads. Discounting early trains buyers to ghost you for a coupon.

One technical guard saves the Page: I store last_inbound_at per conversation and check the window before every send. Under 30 minutes left, nudge three goes short immediately. Window closed, the job exits quietly and waits for the next inbound message instead of firing a 400 against the API.

The table: which surface does what in sales recovery

When I onboard a shop I put this table on one page so the team stops arguing about what the bot should say. Each surface owns one trigger, one reply and one conversion job.

I review this table with owners monthly because it names the leak in one row. Story threads high but handoffs zero means the keyword list is too narrow — usually transliterated Arabic is missing. Handoffs high but recovery zero means nudges are too polite or too late. Threads also stack: a PRICE commenter who replies to a story two days later is merged by Instagram scoped ID, so the score carries over and the second reply opens with "Welcome back — still large?" instead of treating a hot buyer as a stranger.

How I score every auto-reply: hot, warm, cold

My team cannot chase everything, so the bot scores every thread into three buckets and humans touch only hot. Hot means buying intent now: price plus size/color, an address or city, "cash on delivery?", "send link", "احجز", "عايز اطلب", "ابعت اللينك", or YES to a hold. Hot gets stock check plus a human ping within 5 minutes in working hours, and I never dangle discounts on hot — just remove friction and ask for city and phone.

Warm means interested but missing one fact: price-only questions, "available?", emoji-only answers, "ok" / "تمام". Warm stays fully automated through the three nudges, and that is where most revenue is recovered because warm is 60-70% of threads. Cold is the rest: reaction-only, friend tags, or silence past 48 hours with the window closed. Cold gets no more DMs — broadcasts only. I learned this by burning a test Page in 2025: three extra nudges per cold thread tripled blocks in a week.

The score lives as a badge on the conversation and re-computes after every inbound message. One message ("خلاص هاخدها ابعتلي فودافون كاش") flips cold to hot and reroutes instantly. No 20-label taxonomy nobody maintains.

My numbered setup: story + comment-to-DM + 24h window in one afternoon

I set this up for shops in about three hours. Do it in this order — each step depends on the last.

1. Connect Instagram with the right permissions. Link the business Instagram through the Facebook Page and approve instagram_basic, instagram_manage_messages, pages_show_list and pages_messaging with Advanced Access. Send one real DM before touching automation.
2. Route all three surfaces into one inbox. Enable webhooks for story replies, mentions, comments and DMs into the same team inbox, tagged source=story, source=comment_dm or source=dm. AI owns all three by default; humans watch hot only.
3. Build the keyword list in two languages plus transliteration. Start with my list above, add product nicknames ("طرحة / hijab", "مقاس 42 / اوريجينال"), then add the three most common unmatched comments each week. Ten minutes a week keeps it sharp.
4. Write three templates with context variables. Story reply uses story caption plus stock. Comment DM uses post caption plus price plus one question. Nudges use product plus hold language. Every template ends with exactly one question — two questions halve reply rates in my tests.
5. Set the timers and stop rules. Nudges at 60 minutes, 7 hours and 21 hours after my last outbound, only while last_inbound_at is inside 24 hours. Stop on any inbound message, opt-out ("stop", "بس", "خلاص مش عايز") or human takeover, and log every stop.
6. Turn on scoring and route hot to humans. Address or payment intent equals hot with an owner phone ping. Price-only equals warm with AI. Reaction-only equals cold, archived from the DM queue.
7. Launch on five posts, then expand. Enable comment-to-DM on your five best Reels first. If blocks pass 0.5%, narrow keywords. If handoff is under 10% of keyword comments, widen them. Then roll out to stories and the full grid. Check limits first on OT1-Pro pricing.

Arabic + English mirror: why one language cuts replies in half

Half my buyers write Egyptian Arabic, a quarter write English, and a quarter mix both mid-thread: "Price كام؟ عندكم شحن للمنصورة؟" Answer in English only and the Arabic half goes quiet. I detect per message and mirror the same mix.

"بكام الشحن للجيزة؟" gets Egyptian Arabic: "الشحن للجيزة 60 جنيه وبيوصل في يومين. أثبتلك القطعة الكبيرة؟ ابعت رقمك وعنوانك." "Do you ship to Riyadh?" gets English with the Riyadh fee. Mixed gets mixed: "Yes available in large — الشحن للمنصورة 70 جنيه. تحب أثبتهالك بدفع عند الاستلام؟" I always include cash-on-delivery wording for Egypt ("دفع عند الاستلام") because it lifts closes, keep product nicknames the way buyers say them ("الطقم البيج", not dictionary Arabic), and keep quick replies bilingual ("Send address / ابعت العنوان"). One tap beats ten typed words. And I never answer عامية with formal فصحى — a news-anchor tone from a bot drops trust fast.

Escalation only on buying signals: keeping humans out of the noise

I escalate on money, not emotion. "I am angry" alone does not page a human. An address, a payment question or "pack it / احجز / هاخدها" does. That distinction saves my team hours daily.

Human triggers: full address or phone sent, payment link / Vodafone Cash / Instapay asked, invoice requested, or the same stock question asked twice (the bot already failed once). Those reach a human within 5 minutes with the full thread. Everything else — price, size, delivery time, story reactions, friendly haggling — stays automated, and the bot answers in seconds at 2am when humans are at their worst.

One exception runs on risk: damaged-item photos, wrong-order claims, chargeback mentions or medical sizing questions go straight to a human with an apology draft attached. The bot drafts, the human sends, and the bot never promises refunds on its own. My team touches 15-20% of threads instead of drowning in 100%. For how this staffing model compares on price, my teardown is OT1-Pro vs WATI.

Pricing math: $79/mo flat vs ManyChat per-contact pricing

This decides whether comment-to-DM is profitable. Per-contact tools charge for the audience this playbook builds: a shop adding 2,000-5,000 DM contacts a month slides from a $15 starter into $65, then $115, then $200+ as contacts accumulate — before AI add-ons and seats. Your best discovery month becomes your most expensive bill. I watched a cosmetics shop at 28,000 stored contacts pay more for the bot than for ads that month. I paid that bill myself before switching.

I run OT1-Pro at a flat $79/mo for this workload: tracked comments, all three surfaces, scoring, bilingual replies and the 24-hour nudges in one plan. Month one costs slightly more than a starter bot; by month three you pay about half and by month six about a third, because contacts grow and the bill does not. Flat buys freedom to convert every "PRICE" comment without mental math about whether that DM is worth two cents.

Start here: OT1-Pro pricing, then create your free account. Connect Instagram, enable five posts for comment-to-DM, switch on the three nudges, and count recovered threads in week one before paying anything. Stories tonight, keywords tomorrow, nudges the day after — three surfaces, one week, and Instagram stops leaking the sales you already earned.

Stop losing the leads you already earned

OT1-Pro runs your follow-up, broadcasts, and AI replies in one inbox — WhatsApp, Instagram, Messenger, Telegram, and email, in Arabic or English, scored by lead quality, with every AI credit receipted in a transparent ledger. Free plan, no credit card.

Start free → · Sales follow-up automation · Lead follow-up software · Pricing · vs WATI · Talk to the founder on WhatsApp