If you sell on Instagram and WhatsApp in Egypt, you already know the problem. Your ads run through the day. Your Reels get engagement in the evening. And between 9pm and 1am, hundreds of "بكام؟", "متاح المقاس ده؟" and "الشحن كام للاسكندرية؟" messages pile up while your team is asleep. By the time you reply at 10am, half the customers have gone quiet, and a chunk of them bought from your competitor overnight.

This is the operations playbook that MENA storefronts use to handle 500+ WhatsApp and Instagram messages per day, capture the overnight leads that would have been lost, and turn their conversation flow into a real sales machine. No fluff, no theory — just the exact team structure, the routing rules, the AI settings, and the stack.

Why the 11pm-to-1am window matters more than any ad campaign

Meta's own data shows MENA users are most active on Instagram and WhatsApp between 8pm and midnight. In Egypt specifically, evening messaging peaks between 10pm and 12:30am. Yet almost every small storefront runs their reply team from 10am to 6pm — the exact hours when customers are least active.

The math is brutal: if your ad spend produces 200 DMs per day and 60% of them arrive after 8pm, you are effectively burning ad budget on 120 leads that will sit unread for 12+ hours. Studies of MENA D2C brands show that lead-to-sale conversion drops by roughly 60% when response time exceeds 2 hours. Your ROAS is not really an ad problem — it's a reply-time problem.

The team structure that actually scales

Ignore the advice that says "hire 3 more agents." That works for enterprise. For a growing MENA storefront doing $5k–$50k/month in revenue, here's what actually scales:

Tier 1: AI auto-reply (handles 60-70% of messages)

Most inbound messages are simple, repeatable questions: "Do you have this in size M?", "How much is shipping to Alexandria?", "Are you open on Friday?", "Do you accept cash on delivery?". An AI agent trained on your product catalogue, pricing rules, and shipping zones can answer these instantly in the customer's language. If the AI is confident, it replies immediately. If it's not confident (unusual question, complaint, high-value request), it silently escalates.

Tier 2: Human agent (handles 25-35% of messages)

Your human agent focuses on high-value conversations — bulk orders, custom requests, complaints, and any message the AI flagged for review. Because tier 1 has already filtered out the noise, one human agent can effectively handle the volume that used to require three.

Tier 3: Founder / senior escalation (handles 3-5% of messages)

Refunds above a threshold, angry customers, potential viral complaints, and B2B/wholesale inquiries route to the founder or senior manager. This is where you protect your brand and close your biggest deals personally.

The routing rules that make this work

The magic isn't the AI — it's the routing logic that decides what goes where. Here's what we recommend as a starting point for a MENA D2C store using OT1-Pro:

Message type	Signals	Route to
Product availability	Contains size/colour keywords + product name	AI (instant reply)
Shipping cost query	Contains city name + "شحن" / "delivery"	AI (instant reply, uses shipping table)
Order status	Contains order number pattern	AI (looks up in Shopify/Salla/Zid)
Bulk / wholesale	Contains "كمية كبيرة" / "wholesale" / >10 units	Founder
Complaint	Sentiment: negative	Human agent (immediate)
Refund request	Contains "استرجاع" / "refund"	Human agent, escalate if over threshold
Unclear / new topic	AI confidence below 70%	Human agent

Response templates that convert (in Egyptian Arabic)

Generic auto-replies kill conversions. These templates are proven on real MENA storefronts:

Availability + upsell template

أيوة يا فندم، ده متاح في المقاس اللي انتي عايزاه 👗
سعره ٤٥٠ جنيه + ٥٠ جنيه شحن للقاهرة (٧٠ جنيه لباقي المحافظات).
لو ضيفتي معاه أي حاجة تانية الشحن يبقى ببلاش 🎁
عايزة أحفظلك المقاس ده وأرسل الطلب؟

What makes this work: confirms availability instantly, gives price transparently, offers a soft upsell (free shipping over minimum), and closes with a soft-commitment question — not a hard "buy now" ask.

Overnight message template (auto-fires after 10pm)

مرحبا 👋 استلمنا رسالتك.
فريقنا هيرد عليكي بالتفصيل الصبح الساعة ٩، بس علشان مايفوتنيش سؤالك، لو حابة تكتبي المقاس اللي عايزاه والمدينة، هجاوبك مع أول رد بكل التفاصيل 💛

What makes this work: acknowledges immediately (customer feels heard), sets expectation honestly (9am reply), and captures qualifying info while the customer is still thinking about the product — so the morning reply is a closer, not a discovery call.

The stack: what actually goes into the operation

You need three layers, working together:

Layer 1: Unified inbox

All your WhatsApp Business, Instagram DMs, Facebook Messenger, and Telegram messages arrive in one shared inbox. Your team never has to leave one screen. This alone typically doubles reply speed because agents stop context-switching between Meta Business Suite, WhatsApp Web, and Direct.

Layer 2: AI sales agent with your product catalogue

The AI needs to know your products, prices, sizes, shipping zones, and return policy. In OT1-Pro this is configured in your team's AI settings — you paste in your product list and policy notes, and the AI uses that as its knowledge base. Better AI agents (Claude-family in our case) handle Egyptian Arabic dialect natively, so you're not writing scripts in Modern Standard Arabic that no customer actually uses.

Layer 3: Store integration (Shopify / Salla / Zid)

When a customer asks about an order status, the AI should look it up automatically — not tell them "please provide your order number" like a robot. Webhook-based integration with your store platform closes this loop.

What this looks like in numbers

Typical results from MENA D2C brands running this playbook for 30-60 days:

Reply time drops from 4-12 hours to under 90 seconds for AI-handled messages, under 15 minutes for human-handled ones.
Overnight (10pm-9am) leads captured jumps 40-70% because the AI is replying while the team sleeps.
Human agent capacity increases 2-3x because they only handle the messages that actually need judgment.
Total inbound volume goes UP (not down) because customers who used to give up start engaging more when they get fast replies.

Common mistakes MENA stores make when setting this up

Training the AI in Modern Standard Arabic. Customers write in dialect. If the AI replies in classical Arabic, it feels like talking to a government office. Train it to match the customer's tone.
Trying to automate 100% of messages on day one. Start with the easy 40% (availability, price, shipping cost), watch it for a week, then expand. Overreaching creates AI errors that damage trust.
Ignoring Instagram Comments. Instagram comments are messages too — the customer just chose to shout in public. Auto-DM the commenter to move the conversation private. Comment-to-DM automation is one of the highest-ROI features you can enable.
Skipping the overnight auto-reply. The single template above ("we got your message, here's what to send to speed things up") captures 30%+ more overnight qualification data.
Not measuring reply time by hour. Look at your reply-time chart by hour of day. The 10pm-1am window is where most storefronts leak — measure it, fix it.

FAQ

Can I really run this with just 1 human agent + AI?

For $5k–$50k/month revenue stores, yes — usually one human agent handling 80-150 human-touched conversations per day, with AI handling the other 300-500 messages. Above $100k/month you typically need 2-3 human agents plus the AI. Below $5k/month, the founder handles both roles.

Does the AI actually understand Egyptian dialect or just guess?

OT1-Pro routes AI replies through Anthropic Claude via our NaraRouter gateway. Claude handles Egyptian, Gulf, and Levantine Arabic dialects natively — including code-switching ("عايز الأبيض medium please") and misspellings. It also matches the customer's tone: formal Arabic gets formal replies, chat-style dialect gets chat-style dialect.

How much does this stack cost?

OT1-Pro Starter plan is $29/month and covers 3 pages, 500 AI responses/month, and all 4 channels. For most $5k–$50k/month stores that's the right starting tier. As you grow past 2,000 AI responses/month, the Pro plan at $79/month adds bulk campaigns and priority support.

What if the AI replies incorrectly and I lose a sale?

Every AI reply is logged and reviewable. When you spot an incorrect answer, you correct the AI's knowledge base once — and it never makes that mistake again. In practice, human agents miss more messages entirely (because they're asleep) than the AI gets wrong when replying. The comparison isn't "AI vs perfect human" — it's "AI vs no reply at all."

Do I need to be technical to set this up?

No. Setup takes 30-60 minutes: connect your WhatsApp / Instagram / Messenger accounts, paste in your product list and shipping table, tweak the AI's tone. If you get stuck, message the founder on WhatsApp — this is a small enough tool that you talk to a real person.

Try OT1-Pro Free — WhatsApp + Instagram + Messenger + Telegram in One Inbox

OT1-Pro is the unified inbox and AI sales agent built for MENA storefronts. Native Egyptian Arabic AI, per-seat pricing that stays sane as you grow, and a real free tier so you can try it with real customer messages before you pay anything.

Start free → · Pricing · For ecommerce · WhatsApp inbox · Talk to founder on WhatsApp

How Egyptian Ecommerce Stores Handle 500 WhatsApp Orders a Day Without Losing Sales