Skip to main content

AI Brand Voice

Setting up AI Brand Voice in Tattle for all your AI replies

While Tattle’s default AI already drafts replies to guest feedback and online reviews, the AI Brand Voice lets you decide how those drafts sound: how formal, how warm, how apologetic, which words to use, and which to leave out. You set it up once, and every AI reply your team generates from then on follows it, at every location.

This guide walks you through setup step by step, then covers tips for getting the voice right and answers to common questions.

Before you start

  • Who can set it up: Merchant Admins.

  • Where: On a desktop or laptop browser. Setup isn’t available on mobile or tablet.

  • Time: About 5 minutes. Every setting starts with a default, so you only change what you want to.

  • Languages: English and Spanish (other languages coming soon).

  • Cost: Brand Voice is included in your Tattle subscription at no extra cost.

  • Online review replies: Brand Voice applies to online review replies if your account includes Reputation Management. Without it, you can still complete setup, and your voice applies to Tattle Survey replies.

What Brand Voice changes (and what it doesn’t)

Brand Voice changes how AI drafts sound. It doesn’t change whether replies get sent. Your team still clicks Generate AI Reply, reads the draft, edits it if needed, and sends it, the same way you do today.

It covers two kinds of replies:

  • Tattle Survey replies: private email replies sent directly to one guest.

  • Online Review replies: public replies posted on review platforms (requires Reputation Management).

One setting covers both. When you turn Brand Voice on, it applies to every AI reply you generate on either one. You can always edit it at anytime, and saved changes will go live instantly. AI replies for Contact Forms are coming soon.

Step 1: Open Brand Voice

  1. Go to Account Setup.

  2. Select Artificial Intelligence, then Brand Voice.

  3. Click Create a Brand Voice.

The guided setup opens. Step indicators at the top show where you are, and you can go back to an earlier step without losing your work.

Step 2: Set your five voice dials

Each dial asks one question and gives you five options. The default is already selected, and a sample reply updates as you click so you can see what each option sounds like before you pick it.

The five dials include:

  • Formality

  • Tone & Humor

  • Empathy

  • Punctuation & Emojis

  • Apology Stance

You can only choose one option for each dial, and you can reference the sample on the right to understand what each setting sounds like. If you're in between two options, simply pick one to start, as you can always change it later.

Step 3: Add your language and keywords

This step teaches the AI your vocabulary. It has two tabs.

Word Swap

Tell the AI which word to use in place of another. For example: instead of customer, use guest. Or instead of staff, use team members.

  • Enter the word to replace in Instead of and your word in Use this.

  • Fill in both sides of every row. A row with one side empty can’t be saved.

  • Add up to 3 swaps, with up to 20 characters in each field.

  • The first swap (customer → guest) is filled in for you. Keep it, change it, or remove it.

Words To Avoid

List words or phrases the AI should not use, such as “no worries” or “sorry about that.”

  • Press Enter after each entry to add it. An entry you type but don’t add with Enter isn’t saved.

  • Add up to 3 entries, with up to 20 characters each. Multi-word phrases work.

Step 4: Set your rules

Brand DNA

This is where your brand’s story goes. Add short entries under Core Values, Recurring Narratives, and Brand Promises. For example:

  • “We started as a late-night street food cart in 2018.”

  • “Our dough is made fresh daily.”

You can enter up to 3 sentences, maximum 100 characters for each entry.

Guardrails

Guardrails are the rules that keep replies safe. You’ll see two parts here.

Tattle’s built-in protections. These six are always on, and they can’t be edited or turned off:

  • FTC compliance

  • Unauthorized compensation

  • Legal liability

  • Health and safety

  • Spam and sarcasm

  • Inappropriate content

Among other things, the AI is instructed not to admit liability or promise specific compensation, and to give spam and sarcastic reviews a polite reply that doesn’t engage.

Your own rules. Add up to 3 rules of your own, with up to 100 characters each. For example: “Do not offer food safety advice. Direct the guest to contact us.” Your rules add to Tattle’s protections. They can’t remove or loosen them.

Step 5: Add your online review contact and signature (optional)

These two settings only affect Online Review replies. Survey replies ignore them.

  • Contact email: answers “Which email should the AI responses direct guests to?” When a public reply needs to point a guest somewhere, it uses this email. Leave it blank and replies won’t include one.

  • Online Review Signature: how the AI signs off each online review reply, for example “Hot Pizza Team.” You can also enter a tagline, or include emojis here.

Step 6: Test your brand voice

Before you turn anything on, preview how your voice handles real situations. Pick a scenario:

  • Happy 5-Star Guest

  • Mixed 3-Star Guest

  • Vague 2-Star Guest

  • Sarcastic 1-Star Guest

  • Threatening 1-Star Guest

  • Create Custom Test (for online reviews only): paste in a review of your own

Tattle drafts a reply in your voice. Switch between the Tattle Survey and Online Review tabs to see each kind of reply. Custom tests preview on the Online Review tab only.

Not quite right? Use the shortcut buttons under the preview to jump back to the setting most likely to need a tweak for that scenario (for example, Apology Stance on the threatening 1-star). You can go back to any setting, and nothing you’ve entered is lost. Then click Regenerate AI to see the new version.

No Reputation Management? The Online Review tab shows information about the add-on instead of a preview. You can still test survey replies and finish setup.

Step 7: Complete and turn it on

Click Complete. You'll see a confirmation step to activate the Brand Voice. Your account uses one voice at a time: either your Brand Voice or the Tattle Default. Turning your voice on switches the Tattle Default off. If you turn your voice off, replies go back to the Tattle Default.

Using Brand Voice day to day

Nothing changes in how your team replies. On a survey response or an online review:

  1. Click Generate AI Reply. The draft comes out in your voice.

  2. Read the draft and edit anything you want.

  3. Send it.

Your voice applies wherever your team generates AI replies, in the dashboard and the mobile app inbox.

Editing or turning off your voice

Go back to Account Setup → Artificial Intelligence → Brand Voice at any time. The overview page shows your voice, whether it’s on, and when it was last changed and by whom.

  • Click Edit to reopen setup with all your settings filled in.

  • Use the toggle to switch between your voice and the Tattle Default.

  • Changes apply to the next reply generated, at every location.

Tips for a voice that sounds like you

  • Start with the defaults, then change one dial at a time. Test after each change so you know what each one does.

  • Test the hard scenarios first. A happy 5-star reply is easy. The sarcastic and threatening 1-star scenarios show whether your Apology Stance and Empathy are right.

  • Replies feel cold? Try moving Punctuation & Emojis up from Muted or Reserved, or Empathy up one step.

  • Replies feel like too much? Bring Punctuation & Emojis or Tone & Humor down a step.

  • Keep Brand DNA short and specific. The AI may work your story into many replies, including 5-star thank-yous. If it shows up too often, cut it down to the details you’d want guests to hear most.

  • Use Word Swap for internal language. If your team says “team members” instead of “staff,” or “operating partners” instead of “managers,” add those swaps so replies match how you talk.

  • Ban the phrases you edit out most. If your managers keep deleting the same phrase from drafts, add it to Words To Avoid, and press Enter to save it.

  • Put escalation rules in Guardrails. Rules like “Direct allergy concerns to our guest care email” belong there.

  • Paste in a real review. Use Create Custom Test with a review your locations actually got to see how your voice handles your guests.

Frequently asked questions

Does Brand Voice reply to guests on its own?

No. Brand Voice changes how the AI drafts replies. A person on your team still generates, reads, and sends every reply.

Can I set up more than one Brand Voices?

You can only set up one Brand Voice for your brand in this version. We're planning to make updates in the future where you can create multiple Brand Voices, and assign them to different groups or locations.

What happens if I never set up a Brand Voice?

Nothing changes. AI replies keep working with the Tattle Default voice. The Tattle Default stays on until you create and turn on your own voice.

Who on my team can set it up?

Merchant Admins, on a desktop browser. Setup isn’t available on mobile or tablet, or for non-admin users.

Does it cost extra?

No. Brand Voice is included in your Tattle subscription. Online Review replies require the Reputation Management add-on.

Can each location or brand have its own voice?

Not yet. Each account has one voice, used at every location.

Can I upload our brand guidelines and have Tattle fill in the voice?

Not today. Setup is done through the guided steps in this article.

What languages does it support?

English and Spanish.

Can I turn off Tattle’s built-in protections?

No. You can add up to 3 rules of your own, but the six built-in protections always stay on.

Will the AI make things up about my restaurant?

The AI is instructed not to invent facts. Because a person reads every reply before it’s sent, your team can catch and fix anything that’s off.

I don’t like a reply it wrote. What should I do?

Edit the draft before you send it. If you see the same issue again and again, update your voice (a dial, Words To Avoid, or Brand DNA) and test again. Also, we're tracking every edit you make to a reply, so rest assured that your edits are helping us train and upgrade the next version of AI brand voice!

Why don’t I see an Online Review preview?

Online Review previews and replies require Reputation Management. Without it, the Online Review tab shows information about the add-on instead.

Why can’t I save my word swap?

Both sides of each row need to be filled in. Complete the row or remove it.

I added a word to avoid, but it’s still showing up. Why?

Sometimes AI might miss an instruction or so, and you can try clicking "Regenerate AI Reply" to give it another go. We'll keep updating the AI Brand Voice so your experience will only keep getting better.

How fast do changes take effect?

Right away. Your changes apply to the next reply anyone on your team generates, at every location.

Did this answer your question?