Free Customer Feedback Chatbot Template: Rate and Reward

The Customer Feedback template is a ready-made conversational survey for online retailers. Instead of emailing a shopper a link to a page of grid-style rating questions, it opens a short chat that asks one thing at a time: how the site looked, how easy it was to find things, how the delivery went, what arrived, and whether they would recommend you. Every question is a tap, the whole survey reads as one continuous conversation, and it closes by offering a discount code in exchange for the shopper's time.

The template ships with sample copy written for a fictional demo store called Super Company Inc. Every line of that copy -- the welcome message, the product list, the fifteen percent discount offer at the end -- is placeholder text you are expected to rewrite before you publish. Nothing in it describes a real Conferbot customer, and the incentive at the end is a suggestion, not a commitment; if you do not want to give a discount, delete that step and close with a plain thank-you instead.

What makes this template worth starting from is its shape. It is deliberately front-loaded with taps and back-loaded with typing. The first seven questions are rating scales, which means a shopper can get most of the way through without using a keyboard. Only two questions ask for typed input, and both come after the shopper has already invested a few taps and is more likely to finish. That ordering is the single most transferable idea in the template: if you want completed surveys, ask for effort late.

It is built entirely in Conferbot's no-code chatbot builder, so you edit it by clicking nodes on a canvas and retyping their text. You can see the rest of the category on the surveys templates page, and if you would rather measure a single loyalty number than a full experience breakdown, the NPS survey template is the narrower tool for that job.

This template fits any business where a customer places an order online, waits for it to arrive, and then unwraps it -- because those are exactly the three moments it asks about. Direct-to-consumer brands, marketplaces, grocery and food delivery, electronics and accessories retailers, and subscription box operators all map cleanly onto its structure.

It suits you particularly well if your feedback problem is coverage rather than depth. A long-form survey tends to be answered by the two extremes: delighted customers and furious ones. A tap-only conversation that takes under a minute pulls in the quiet middle as well, which is where most of your revenue actually lives. Because the template separates store experience from delivery experience from product condition, the results tell you which of the three is dragging, rather than leaving you with one blended satisfaction figure you cannot act on.

There are two cases where you should reach for something else. If you need to diagnose a specific feature or release rather than a whole purchase journey, the product feedback template asks better questions. And if what you actually want is to route unhappy customers to a person rather than to collect scores, look at the support and FAQ templates, which are built around handover rather than measurement. This template has no human-handover node in it -- it is a collection flow, and everything it gathers lands in your responses table rather than in front of an agent.

Here is the actual flow as it ships, in the order a shopper experiences it. The wording below is the template's own sample copy.

1. The opening and the opt-in

The bot opens with a welcome-node: "Thanks for shopping with Super Company Inc." It then uses an n-choices node to ask permission before asking anything else -- "We'd love to know a bit more about your experience with us." -- with a single button reading "Let's share some feedback." Offering one button rather than a yes-or-no pair is a deliberate choice: it frames participation as the default path while still requiring an explicit tap, so the shopper starts the survey having already agreed to take it.

2. Rating the online store: four questions

A message-node sets up the first block -- "How would you rate our online store, based on:" -- and then four rating-choice nodes fire in sequence. The first, "Design of the site," uses the smiley rating type, so the shopper taps a face rather than a number. The next three use a five-point scale: "The range of products," "Finding what you needed," and "Product descriptions." Answers are stored as design rating, range rating, finding rating and descriptions rating.

The split is worth noting. Site design is subjective and emotional, so a smiley scale is the honest instrument for it. Product range, findability and description quality are judgments, so they get numbers. Mixing scale types inside one survey is normally a mistake, but here it maps to a genuine difference in what is being asked.

3. Rating the delivery: three questions

Another message-node marks the transition -- "Awesome, thanks! Now let's score our delivery..." -- which matters more than it looks, because it tells the shopper the topic has changed and roughly how much is left. Three more five-point rating-choice nodes follow: "How would you rate the delivery time?", "The friendliness of the delivery driver?", and "And the quality and condition of the received product?" These are saved as delivery time rating, driver friendliness rating and quality condition rating.

Separating driver friendliness from delivery time is the kind of detail that makes results actionable. A poor delivery score that turns out to be entirely about timing points at your logistics partner's scheduling; one that is about the driver points at their training. A single "how was delivery" question cannot tell those apart.

4. What did they order?

An n-select-option node asks "Could you tell us from what did you order?" and offers four buttons: Fries, Tablet, Footwear and USB. The answer is stored as order. This list is plainly demo filler -- it is a deliberately unrelated set of items chosen to show the node working, not a catalog -- and it is the first thing you should replace with your own product categories. Capturing the category is what lets you segment every rating above it: the delivery scores for footwear and for fries are not the same question.

5. The recommendation score and the open question

One more five-point rating-choice node asks "Thanks, and would you recommend us to your friends?", stored as recommend rating. Placing the recommendation question after all seven specific ratings is good survey design: by this point the shopper has already thought concretely about the site, the delivery and the product, so their recommendation score is an informed summary rather than a first impression.

Then an ask-custom-question node opens the only free-text box in the survey: "Now, do you have any suggestions for us?" It is the twelfth step, not the first, which is exactly why people answer it.

6. The incentive and the close

A message-node makes the offer -- "Finally, would you like 15% off your next shop?" -- and an ask-email node collects the address: "Drop us your email address if you want us to send you a coupon." That node validates the input and replies "Please enter a valid email address" if what it gets is not one, so you do not end up with a list of typos. A closing message-node confirms: "Thank you, you will receive the coupon code shortly."

Note what this arrangement does. The email is requested last, after the shopper has answered everything, and it is framed as the delivery address for something they want rather than as a data field. If they skip it you still keep every rating. Note also the commitment it creates: whatever offer you put in that message, you have to actually send. Wire the fulfilment before you publish.

Seventeen nodes make up this flow, drawn from six Conferbot node types. Understanding which type does what is the difference between rewording the template and genuinely reshaping it.

Every rating-choice node writes to a named answer variable, which is what turns a conversation into a dataset. Because the variables are named per question rather than pooled, you can chart driver friendliness on its own, cross it against the product category, or export the whole set and look at which rating moves the recommendation score most. The named variables are also what you reference when you pass answers onward to another system.

The flow is strictly linear -- there is no branching in it. Each rating node routes every one of its five possible answers to the same next node, which is why a one-star answer and a five-star answer produce an identical conversation. That is a reasonable default for a survey whose job is measurement, and it is also the most obvious place to extend the template, which the customisation section below covers.

Six steps take this from template to published bot. None of them require code.

Step 1: Open the template and make it yours

Find Customer Feedback in the Surveys category of the template library and use it to create a new bot. You are working on your own copy from that point on, so you can rewrite anything without affecting the original.

Step 2: Replace every piece of demo copy

Work left to right across the canvas. Swap Super Company Inc. for your brand name in the welcome-node. Replace the four product options -- Fries, Tablet, Footwear, USB -- with your real categories, adding or removing buttons as needed. Decide what the incentive message actually says: your real discount, a free-shipping code, an entry into a draw, or nothing at all. If you drop the incentive, delete the ask-email node with it rather than asking for an address you have no reason to want.

Step 3: Prune the questions you will not act on

Eight scored questions is on the long side for a post-purchase survey. Be honest about which results would change a decision. If you do not use your own couriers, "the friendliness of the delivery driver" may still be worth keeping as evidence for your logistics partner -- but if nobody will ever read it, cut the node and shorten the survey. A five-question survey that gets finished beats an eight-question one that gets abandoned.

Step 4: Wire up where the answers go

Connect the destinations for your responses. Conferbot has sixteen native integrations including Google Sheets for a running response log, Gmail for per-response notifications, Slack for alerting a channel, HubSpot or Zoho CRM for writing feedback onto the customer record, and Airtable or Notion if you triage suggestions there. Anything beyond that list is reachable through Zapier, which connects to over two thousand further apps, or through webhooks and the API on Starter and above. The integrations hub has the full set.

Step 5: Pick your channels

All ten channels are available on every plan, including Free, so this is a question of where your customers already are rather than what you are paying. Details on each are on the omnichannel page.

Step 6: Test the whole path, then publish

Run it end to end at least twice. Check that the smiley scale renders the way you expect on a phone, that a deliberately malformed email triggers the re-prompt, that your integration receives all eight variables under the names you expect, and that the coupon you promised actually arrives. Then publish. You can watch completion and drop-off per question in chatbot analytics and read individual transcripts to see where people stalled.

The template is a skeleton with good bones. Four changes are worth considering once the basic version is live.

Branch on low scores

Because every rating currently routes to the same next node, a shopper who rates their delivery one out of five gets the same cheerful "Awesome, thanks!" as one who rated it five. Rewiring the low answers of a rating node to a different path is straightforward on the canvas, and it is the highest-value edit available here. Send one-and-two-star answers to an ask-custom-question node that asks what went wrong, or to a human-handover node so an agent picks the conversation up live. Live chat and the unified inbox are on every plan including Free, so adding a handover costs nothing; see live chat for how the takeover works.

Ask the product category first

The n-select-option node currently sits between the delivery ratings and the recommendation question. Moving it to the front, just after the opt-in, lets you branch the whole survey on what was bought -- perishables get questions about freshness and packaging, electronics get questions about setup and instructions. Same node, different position, considerably more useful data.

Cut the typing entirely

For a pure tap survey, delete the ask-custom-question and ask-email nodes and close straight after the recommendation score. You lose the qualitative answers and the coupon hook, but you end up with a survey that takes about twenty seconds, which is the right trade if you are asking after every single order rather than occasionally.

Add conditional thanks

A small touch: route high recommendation scores to a message-node that asks for a public review, and everything else to a plain thank-you. You are asking for advocacy only from the people who just told you they would give it.

Which channel to run it on

A post-purchase survey works best where the order conversation already happened. If you confirm orders and shipping on WhatsApp, run the survey there too -- it lands in an open thread the customer already trusts, and rating buttons render natively; see WhatsApp chatbots. If your traffic is web-first, the website widget triggered on the order confirmation page catches people at the moment the purchase is freshest. Instagram and Messenger suit brands whose customers arrived through social in the first place. Telegram, Slack, Discord, LINE, Microsoft Teams and the Mobile SDK round out the ten available channels, and one bot can serve several of them at once.

One scheduling note for 2026: this survey asks about delivery and product condition, so it has to fire after the parcel arrives, not after checkout. Trigger it on your delivery-confirmation event rather than your order-confirmation one, or you will be asking people to rate a box they have not opened.