Auto-answer ‘price koto?’ and ‘details plz’ from your catalog

You post a new batch of cotton kurtis at 8 pm and boost it for ৳1,000. By 9 pm the inbox has sixty new chats. Fifty of them say some version of "price?", "dam koto", "details plz" or just "pp". Someone on your team copies the same price, pastes it, scrolls, pastes again, and gets three of them wrong because two kurtis look alike.

These messages are not hard questions. They are the same question, asked in thirty different ways, about different products. That is exactly why fixed auto replies fail on them, and why a catalog-based reply works.

Here is how to answer them automatically, with the right price every time.

How do customers actually ask for the price?

Customers ask for the price in Bangla, Banglish, English and shorthand, often without naming the product. The table lists the variants a Facebook shop sees in a normal week, and what each one really needs.

Two patterns stand out. First, most messages don't name the product. Second, "details" almost always means more than the price. A good automatic answer handles both.

Why can't a keyword auto reply answer "price koto"?

A keyword auto reply sends one fixed text when a word matches, so it can't quote different prices for different products. It works if you sell one item. It falls apart with fifty.

It also misses spellings. A rule on "price" never sees "dam koto", "koto taka" or "দাম কত". You can add more keywords, but you will still be sending the same text to the person asking about a ৳650 scarf and the one asking about a ৳3,200 saree.

Keyword replies are still useful for fixed facts that don't depend on the product: your delivery charges, your shop address, your exchange policy. Keep them for that.

How does a catalog-based reply work?

A catalog-based reply looks up the product the customer means in your product list, then answers with that product's listed price, options and stock. The bot is a lookup with good manners, not a writer of prices.

In practice the reply goes through three steps:

1. Work out the product. From a code, a product name, a colour or the earlier chat.
2. Look it up. Read the current price, any old (compare-at) price, the sizes or colours and the stock.
3. Answer and move forward. Give the price, answer anything else asked, and offer the next step.

The exchange should look like this:

Customer: KR-114 er dam koto? M size ache?

Page: KR-114 cotton kurti ta 1,190 taka (agey chilo 1,450). M size ache, shudhu 3 ta baki. Dhakar bhitore delivery 60 taka. Order korben?

All prices there are examples. Every number in a real reply must come from your list.

What should the bot do when the customer doesn't say which product?

When the product is unclear, the bot should ask one short question instead of guessing. A wrong price costs more than one extra message.

Good clarifying replies are short and give the customer an easy answer:

Kon product ta janben? Post er niche code ta (jemon KR-114) pathale ekhuni dam bole dicchi.

কোন প্রোডাক্টটার দাম জানতে চান? পোস্টের কোডটা পাঠালে সাথে সাথে বলে দিচ্ছি।

The best fix happens before the message arrives. Print a code in every caption and on every product photo: "Code: KR-114". Customers start typing the code because it is right there, and every lookup becomes exact.

Screenshots are the other common case. A customer sends a photo of your post with "eta koto?". Many chatbots, SaleMag's included, read text, not images, so that chat should go to a person, or the bot can reply asking for the code. Tell your team to watch for these during a campaign.

How do you prepare your catalog so answers are right?

Answers are only as good as the product list behind them. Before you switch anything on, check every product you are advertising against this list.

Write names the way customers talk about the product. If buyers call it a "three piece", don't only call it "unstitched salwar suit set". If you sell in Bangla, put the words customers use into the description or FAQ.

How do you stop the bot from quoting a wrong price?

You stop wrong prices by making the bot read prices instead of write them, and by checking every amount it sends. Three rules cover it:

• Look up before quoting. The bot must search the catalog every time it states a price, stock level or option.
• Guard every amount. Any money figure in a reply that isn't a real catalog price should be caught and replaced.
• Totals come from the system. For two pieces plus delivery, the total should be calculated, not composed by the AI.

Also decide what the bot should never do: offer discounts you didn't set, promise delivery dates you didn't write down, or quote a product you don't stock. Put those rules in its instructions in plain words, such as "Do not offer discounts" or "If the product is not in the list, say we don't have it".

How does SaleMag answer price questions?

SaleMag's AI chatbot answers price questions on Messenger, WhatsApp and Instagram from your knowledge base, which can include your whole product catalog, synced from Shopify, WooCommerce or BigCommerce or added by hand. It re-indexes when products change, replies in the customer's language including Banglish, and is instructed never to invent prices.

With AI order taking on, it goes further. It must search the catalog by name, SKU, category or brand before stating any price, stock level or option. A price guard catches any amount in its reply that isn't a real catalog price, and the server calculates totals. It can send up to three product photos, build the cart, and send an order form. Bulk orders and complaints go to your team, and the bot stays quiet for 15 minutes after a teammate replies.

For the wider setup, including Meta's own instant replies and what they can't do, read how to set up Facebook Page auto reply.

Next steps

Pick your next boosted post. Add a code to the caption, check the price and options in your catalog, and paste twenty real "price?" messages from your inbox into a test chat before the ad goes live.

Start free with SaleMag to connect your Page and catalog, or see pricing.