Before activating Shopper for your subscribers, you'll want to see how it handles real conversations. Postscript gives you two ways to test: Playground (inside the app) and a live SMS demo. Both let you verify that Shopper sounds like your brand, recommends the right products, and answers questions accurately.

Testing is an iterative process. The goal is to have a conversation, spot anything that feels off, make an adjustment in Brand Center, and test again. Most merchants go through a few rounds of this before they feel confident activating.
​

Playground is an in-app testing environment where you can simulate a Shopper conversation without sending a real SMS or using any credits. You'll find it inside the Shopper settings page as well as in the Brand Centre.
​

To start a test, just type a message the way a subscriber would. Try things like asking about a specific product, asking for a recommendation, asking about your return policy, or saying something vague like "I'm looking for a gift." The more varied your test conversations, the better sense you'll get of how Shopper performs.
​

Playground reflects everything you've configured in Brand Center — your brand voice, Content Library, and Product Catalog. When you make a change in any of those areas, it will resync the information and you can then immediately test the impact in Playground.
​

For a more realistic test, you can have a real text conversation with Shopper on your phone. Go to your Shopper Settings page and either scan the QR code shown there, or text SHOPPERDEMO to your shop's number.
​

This starts a live conversation that looks and feels exactly like what your subscribers will experience.
​

Share the QR code with your team so multiple people can test. The more perspectives you get, the better.
​

Brand voice — Does Shopper sound like your brand? If the tone feels wrong, adjust your Brand Voice & Tone settings in Brand Center. See Setting Up Your Brand Voice & Tone for guidance.
​

Product recommendations — When you ask Shopper to help you find a product, does it recommend items that make sense? Does it ask good discovery questions to narrow down what you're looking for? If the recommendations feel off, check that your Product Catalog is pulling in enough information and that the right products are toggled on. See Managing Your Product Catalog for details.
​

Policy and FAQ answers — Ask Shopper about shipping times, return policies, warranty information, or anything else your subscribers commonly ask about. Compare the answers to what's actually on your website. If something is wrong or missing, update the information on your website or in your Content Library. See Content Library Overview and AI Best Practices for Content Library.
​

Offer behavior — If you've configured a Strategic Offer, try acting like a price-sensitive subscriber. Say something like "that's a bit out of my budget" or "do you have any deals?" and see if Shopper surfaces your offer appropriately. Is it missing some key information about the offer? If something is off, double-check your offer configuration in Setting Up Shopper Offers.
​

Edge cases — Try asking Shopper about those edge cases that are unique to your brand. How does it handle a sensitive question or a topic unrelated to your brand? Note that Shopper is trained to not respond if it detects topics that are potentially spam, offensive, political, sent by mistake, or “I’m driving”.
​

When you find something that needs fixing, the process is straightforward:
​

Identify the issue (wrong answer, awkward tone, bad recommendation). Go to the relevant area of Brand Center or Shopper Settings and make the adjustment. Come back to Playground and test the same scenario again. Repeat until it feels right.
​

This loop is the most valuable part of your setup process. Every round of testing and adjusting makes Shopper smarter and more accurate. Most merchants find that three to five rounds of testing covers their main use cases and gives them confidence to go live.
​

Don't aim for perfection before launch. Shopper will continue learning and improving after activation, and you can always make further adjustments to Brand Center as you observe real subscriber conversations. The goal of pre-launch testing is to make sure the foundations are solid: voice, product knowledge, and policy accuracy.
​

Playground also supports in-app message-level feedback. If a Shopper response feels off, you can click the thumbs-down feedback button. You'll be prompted to select a reason — Emoji usage, Offer details, Brand voice & tone, Product reference, or Other — and can add context to help the team understand the issue.
​

After submitting, you're taken directly to the relevant Brand Center section so you can make the fix yourself on the spot. This feedback is actively tracked and used to drive product improvements, and is laying the groundwork for Shopper to eventually self-correct based on feedback automatically.
​

If you encounter an issue in the Conversations view after going live, the same feedback flow applies. Submit negative feedback in-app before escalating a quality issue or filing a product request or bug ticket — it gives the team the record and context needed to resolve things faster.