Auto-Reply to Size and Color Questions: The "Do You Have It in M?" Flow
You can answer "do you have it in M?" on Instagram automatically, with stock info. We build the flow with the Product Detection, Variant Selection and AI Product Q&A steps.
In this post
When a customer writes "do you have it in M?" in your DMs, you can answer automatically. Your message automation uses three product steps for this: Product Detection finds the product, Variant Selection asks for size and color, and AI Product Q&A answers open questions from the product details.
This guide shows how to put these steps together in one flow.
Before you start: your catalog has to be ready
All of these steps take their answers from your product records. If your product has no size and color variants, the system can't ask about them. If you haven't entered stock, it can't tell the customer whether something is in stock.
You can add products one by one on the My Products page or upload them in bulk from a file. We covered uploading products with variants in the Excel variant product import guide. In short: each variant is its own row and carries its own stock count.
What each step does
| Step | What it does | Outputs |
|---|---|---|
| Product Detection | Looks for the product name or code in the customer's message and shows the products it finds as cards. | Carousel buttons, Not Found |
| Product Context Check | Checks whether a product is already known in the conversation. | Product Found, No Product |
| Variant Selection | Asks about options like size and color in turn until one variant is left. | Resolved, No Match |
| AI Product Q&A | Answers an open question from the product record. | Answered, Unknown |
Variant Selection and AI Product Q&A don't find the product on their own. You can only reach them from a path where the product is already known: the matching branch of Product Detection, or the Product Found branch of Product Context Check. The settings of both steps say this too.
Scenario 1: The customer asks by product name
The customer writes "do you have the cotton t-shirt in M?". Here's how to build the flow:
- Trigger. Open your message automation. Use the Manual Trigger and enter words like "size", "color", "stock" and "in stock". If you also want to catch other phrasings, use the AI Intent Detector and write a description like "The customer is asking about the size, color or stock of a product". The AI Intent Detector takes from your monthly credit every time the AI is asked. Keywords use no credit.
- Product Detection. Connect it after the trigger. The product name or code in the message is searched in your catalog. The product it finds goes to the customer as a card with its photo and name. You can show the price in the card's subtitle. If more than one product is found, the cards sit side by side and the customer picks the right one. Under the card you add Carousel Buttons, for example "Choose size". A card can have up to 3 buttons.
- Variant Selection. Connect the output of the "Choose size" button here. When the customer taps the button, the system asks about the product's options one by one: first "Which Color?", then "Which Size?". Each question goes out with quick reply buttons and only shows values that fit the earlier choice. If the red one has no L, a customer who picked red won't be offered L.
- Resolved branch. Add a Text Message and use the variant variables:
Status of Variant (Size/Color): Variant Availability. Price: Variant Price.
If the customer picked red in M, the message shows something like "Color: Red, Size: M" and the status "In stock" or "Out of stock".
- No Match branch. If the customer types something that isn't on the list instead of tapping a button, the flow goes to this branch. Put a short pointer here, or use the Live Support step to hand the conversation to you.
- Not Found branch. This branch can't be left empty. Add a Text Message that asks for the product name: "Which product are you asking about? Could you type its name or code?" Connect this message's main output back to Product Detection, so the customer's answer is searched again. You also need at least one quick reply on this message, for example "Talk to a person". Connect it to a step other than Product Detection. A customer who can't find the product uses that button to get out of the loop. The panel warns you if this quick reply is missing.
Only a Trigger or a Text Message can connect to Product Detection. Outside of this loop, a Text Message connected to Product Detection can't have quick replies, because the step is waiting for the customer to type.
With this setup, the customer sees a conversation like this:
What sizes does the cotton T-shirt come in?
Which Color?
Red
Which Size?
M
Status of Color: Red, Size: M: In stock. Price: 299.90 ₺.
The skeleton of the flow:
Writing the question yourself
Variant Selection writes the question on its own. The default looks like "Which Size?". If you want a warmer tone, use Axis Questions (custom) in the step's settings to write your own sentence for each option, for example "Which size would you like?". Any option you leave empty uses the default question.
If an option has a lot of values, only the first 13 are shown because of Instagram's quick reply limit.
Scenario 2: The customer came to the DM from a comment
You sent a private message to someone who commented on your post, and your comment automation ends with a Link Message Automation step. The customer replies to that message with "do you have it in M?". In this case the product is already known, because the post they commented on is linked to a product.
In the message automation you linked, use Product Context Check instead of Product Detection:
- On the Product Found branch, go straight to Variant Selection or AI Product Q&A. You don't need to ask the customer for the product again.
- On the No Product branch, add a message that asks for the product name, followed by Product Detection.
Don't use product variables like Product Name on the No Product branch. They will be empty.
Product Detection also recognizes this case. If the message doesn't mention a new product name and a product is already known in the conversation, it uses that product. If the card has a single connected button, it skips the card and continues from that button's branch.
We set up the move from comment to DM step by step in the auto DM for price comments guide.
Scenario 3: Open questions
Not every customer asks about size and color. AI Product Q&A also answers questions like "What's the fabric?", "How long does shipping take?" or "Can I return it?".
This step answers only from the information in your product record: the description, highlights, attributes, variants and their stock status, handling time, return policy, warranty text and similar fields. The AI doesn't make up answers. If the information isn't in the product form, or the AI isn't sure, the flow goes to the Unknown branch. You'd usually connect that branch to Live Support or to a message like "Let me check with the team, we'll get back to you shortly".
One thing to watch: this step doesn't wait for new input. The question it answers is the customer message that arrived when the flow reached the step. So place it where the customer's question flows straight in. For example: Trigger, then Product Context Check, then AI Product Q&A on the Product Found branch. If you put it behind a carousel button, the button's text becomes the question and the answer won't make sense.
AI Product Q&A uses your AI allowance too. The allowance is shared with AI triggers. When it runs out, the step doesn't call the AI at all and goes to the Unknown branch, so the customer still gets a reply.
Getting better answers
- Fill in the product form. Everything AI Product Q&A knows comes from the form. Add the things people often ask about, like fabric, fit and shipping time, to the product.
- Keep variant names consistent. If one product uses "Size" and another uses "Sizing" for the same thing, customers get two different questions. Use the same name for the same option on every product.
- Keep stock up to date. The Variant Availability value comes from the stock count. If you don't lower stock after a sale, the customer will be told "In stock".
- Connect every branch. If the Not Found, No Match or Unknown branches are left empty, the panel shows them as errors and asks you to fix them before the flow can be activated. These branches exist so the customer never goes unanswered.
For other ways to use message automation, see our Instagram DM automation page.
Frequently asked questions
Will a sold-out size be shown to the customer as an option?
Variant Selection asks about all of a product's variants and doesn't filter them by stock. So put the Variant Availability variable in the message on the Resolved branch. The customer sees in that message whether the variant they picked is in stock or out of stock.
Does AI Product Q&A use AI credit?
Yes. Each call is deducted from the monthly allowance you share with AI triggers. If the allowance is used up, the AI isn't called at all and the flow goes to the Unknown branch.
What happens if the customer types the size instead of tapping a button?
Variant Selection compares what the customer typed with the product's option values and ignores upper and lower case. If the answer doesn't match any option, the flow goes to the No Match branch. Tapping the quick reply buttons is the safest way.
