Just because I told you and you sent a random chunk of words with a WhatsApp AI chatbot relating to dresses doesn’t mean they would notice. But presenting it tailored based on their purchase history will definitely make them notice. This blog is all about customer retention using a WhatsApp AI chatbot and a guide on how to implement it.
Scenario:
Retaining a business’s customer base is the answer to the million-dollar question for every business owner. Most businesses take measures to bring in new customers while they keep losing their onboarded customers along the way. As per statistics, in terms of marketing, onboarding or bringing in a new customer costs 5x more when compared to holding on to their past recurring customers. This blog provides one of the many ways to retain their existing customers. And this way of customer retention can be done by a WhatsApp AI chatbot workflow. Let’s dive deep into the technical and analogical aspects of customer retention for a fashion retail store.
The Problem We Are Solving
We are searching for gold mines while we have diamonds sitting idle in our database. While thousands of customers were purchasing retail units from our site, they gave away their numbers, and those numbers are in our database, sitting safe and secure, but untouched. Instead of these assets getting buried under dust, let us put them to their utmost use.
What if we instigate them into increasing their consumerism? A master stroke, right? And how can we achieve this sort of marketing brilliance? 😌
No wonder the answer is our WhatsApp AI chatbot for customer retention.
I know you have already guessed it. Anyway, I am continuing as if you still don’t know the answer. Now, let’s move to further sections to find how to implement this.
Our Customer Retention Flow
For every 4 months in a year, a seasonal sale will be hosted. This is how the flow looks.
Here we can see that the conversion rate through WhatsApp is just 5%. But this is a decent number in terms of marketing. The reason is that here we haven’t used paid ads. Adding to that, there’s no need to worry about email open-rate anxiety, and, more than that, no social media algorithm to fight.
How WhatsApp AI Chatbot Works
Tiers of Customer Segments:
After every sale or sale cycle, your customer list must be split into 3 different tiers, and the tier in which they fall makes a huge difference in terms of marketing.
💎 Tier 1 - Early Access
Customers who fall in these tiers should have purchased twice in the last 4 sales cycles. They should be provided with early sale access before the other tier customers. Buy once you are lucky enough to be Tier 2. Buy twice and BOOM, you are Tier 1.
Customers who signed a pact with loyalty. 😇
Hi {{1}},
You’ve unlocked Tier 1 status! 💎
Because you’ve shopped with us twice in recent sales,
you get EXCLUSIVE EARLY ACCESS to our {{2}} —
before anyone else and VIP-only benefits.
Enjoy a special {{3}}% loyalty reward,
just for our top shoppers.
🛍️ Enter early here: {{4}}
Reply "SHOP" to browse with our style assistant or
"STOP" to unsubscribe.🛍️ Tier 2 - Special Coupon
Customers in tier 2 are those who made at least one bulk purchase in a sale within the past year. These people are our potential Tier 1 consumers, so let’s offer them a coupon with some special discounts.
In other words, impulsive buyers who stood with us. 😊
Hi {{1}},
You’re officially on our radar 👀
Since you made a big purchase recently,
we’ve saved a special {{3}}% coupon
for our {{2}} — just for you.
Let’s make this your Tier 1 moment 😉
🎁 Claim your offer: {{4}}
Reply "SHOP" to chat with our style assistant or
"STOP" to unsubscribe.🔁 Tier 3 - Comeback Offer
These are customers who had a history with us. Not a beef, but a happy purchase history. To bring them back in, let’s give them the fruits of being a part of us, like a comeback offer.
Our once-in-a-blue-moon customers. 😲
Hi {{1}},
We’ve missed you! 🥹
It’s been a while since your last purchase,
so we’re giving you a {{3}}% comeback treat
for our {{2}}.
Let’s make it official again 💛
✨ Come back here: {{4}}
Reply "SHOP" to reconnect with our style assistant or
"STOP" to unsubscribe.The gaps our WhatsApp AI chatbot fills
Since you wouldn’t want your business WhatsApp account to be flooded with something like 300 manual conversations. This is straight up a challenge for any support team to handle.
So, for those who badly want to be on the winning side?
The secret to winning repeat customers? This time, it is a WhatsApp AI chatbot.
Sending a WhatsApp template message regarding the sale.
Providing information about the products
Sharing the website or purchase URL directly via chat.
Storing them in our CRM database as per the customer’s intent on product purchase.
Handing over customer interaction to a human agent.
Technical Implementation:
So, this is the part where most bloggers send goodbyes after marketing their product, but we won’t. Now, let’s have a look at the technical implementation of this stack.
1. Cron Job Scheduler:
Tasks like daily backup, cache clearing, and periodic system maintenance on a regular basis are achieved with the help of schedulers like cron job.
A Cron Job is nothing but a time-based task scheduler that can schedule scripts or commands to run at specific intervals. It can be seen implemented in Unix-based operating systems such as macOS and Linux.
The Cron Job Scheduler is the soul of WhatsApp AI Chatbot automation because we humans forget, but this guy doesn’t.
Consider the cron daemon, the “Mr. Perfect” that goes to work and does all his work with punctuality.
For example,
Including crond: schedule cron jobs.
managing SSH connections.
serving HTTP or nginx to deal with web requests
initiating mysqld to run the database engine
The Crontab is the guy who orders the Cron daemon to do things. Once the Cron daemon is ready to set the floors on fire, Crontab will be its user guide on how to do that.
It hands out things to Cron Daemon, such as a five-field time format (minute, hour, day of month, month, and day of week) and some commands to be carried out at that specific time.
During our sale days, i.e., once in every 4 months, our scheduler initiates this entire workflow.
System Boots Up.
The cron daemon initiates the background system.
Crontab hands out the to-do instructions.
Fetches client information from the database.
Template messages are sent by the WhatsApp AI chatbot using Meta APIs.
2. Database Schema:
No, let’s get into the design of our schema.
CREATE TABLE customers (
id SERIAL PRIMARY KEY,
name VARCHAR(100),
whatsapp_number VARCHAR(20) UNIQUE NOT NULL,
last_purchase_date DATE,
purchased_last_sale BOOLEAN DEFAULT FALSE,
discount_tier VARCHAR(10) DEFAULT 'standard',
opted_out BOOLEAN DEFAULT FALSE,
created_at TIMESTAMP DEFAULT NOW()
);
CREATE TABLE sale_cycles (
id SERIAL PRIMARY KEY,
cycle_name VARCHAR(100),
start_date DATE,
end_date DATE,
vip_discount_code VARCHAR(50),
standard_discount_code VARCHAR(50),
is_active BOOLEAN DEFAULT FALSE
);
CREATE TABLE campaign_logs (
id SERIAL PRIMARY KEY,
customer_id INT REFERENCES customers(id),
cycle_id INT REFERENCES sale_cycles(id),
message_sent_at TIMESTAMP,
replied BOOLEAN DEFAULT FALSE,
purchased BOOLEAN DEFAULT FALSE,
purchase_amount DECIMAL(10,2)
);I know I have flabbergasted you by including these fields specifically and not the other way.
Schema, Dude! 😌
Okay, most fields vary as per your business type, product type, and the process involved in sales. But the fields like whatsapp_number, last_purchase_date, purchased_last_sale, etc., are mandatory to decide the type of tier they fall into.
3. WhatsApp AI ChatBot:
Now that we are into the title track, I have to be more specific. So, leaning into the specifics, the WhatsApp AI chatbot works based on the incoming webhooks. So, what does our WhatsApp AI chatbot do?
When a customer sends a message to our promotional WhatsApp account, it receives a POST webhook request, and that will be shown as an event.
From the event, it extracts data such as customer phone number and other data from the nested JSON payload. This extracted user response will be sent to LLMs trained on RAG models. Based on the trained data and the response from the user, an apt reply will be generated. This will provide an authentic response to the customers and make the conversation live.
4. CRM update with post-purchase:
When a purchase is completed using the link we provided to our customers, let’s tag them as “Batman” in our CRM.
“Not the heroes we deserve, but the heroes we need.”
These customers are the assets we would need to have a tab open and prioritised during our upcoming sales days.
Now let’s rank our heroes. “Come on, we all have our favourites. Don’t we?”
Here we categorise them based on their history of purchases they made.
Saved us once, but long back, Tier 3. 😇
Saved us a season back, nothing short of Tier 2. 😍
Been there more than twice in the past 2 years; definitely Tier 1. 💎
Segmenting them into categories should be a no-brainer, as it makes multiple aspects of marketing easier. Also, using these data for our WhatsApp AI chatbot.
The Customer Retention Psychology
WhatsApp has an open rate of 98%, whereas email has just 20%. While our promotional emails are left to rot among other promotional emails, WhatsApp doesn’t deal with that. The reason for this is that WhatsApp promotional messages roam around personal messages from our friends and relatives.
It doesn’t compete with 67 other promotional messages like emails.
Now that we have got their attention, we can say
“We value you because you have been with us. Thank you for that. Here is a coupon code for you.”
This is a
“I am instigating you to buy from us again.”
kind of message, and the good thing is they are kind enough to buy from us again. This is because of a reward loop that we induced using our 3-tier discount system.
Customers who had always been a part would want to contribute more, and customers who haven’t will feel the mild sting of exclusion.
Keep A Track Of
Keep track of the data collected during each sale period to improve the workflow. The message-to-enquiry rate in subsequent sales will tell the story about the retention strategy. A growth in the message-to-enquiry rate strongly suggests that our retention strategy is high-time efficient. Even though the message-to-enquiry rate (i.e., 5%) seems like a small number in sales, it provides stability to our business.
The Easier Way: Agentzee
After scrolling through the entire blog, I can sense your slothfulness. As a fashion retailer, you wouldn’t want your mind to be boggling with this much technical stuff. “I know it is a lot to take in, dude.”
“But one way or the other, you have to implement this, right?” So, let me provide you with an easier way, and that’s us: Agentzee. (Agentzee)
We not only provide a WhatsApp AI chatbot but also one for Instagram and Facebook. Also, adding to that, we got a voice call campaign and thirty more third-party platform integrations.
“It means I have got it all covered under one roof. Try signing up.”
Here, you can explore how it works:
Agentzee WhatsApp AI chatbot