- Bruxel.AI Blog
- Brand voice in product pages: how to write the same way across thousands of products
Brand voice in product pages: how to write the same way across thousands of products
By Manuel Mattar3 min read
In short
Brand voice in a catalog is the set of rules that makes thousands of product records sound like the same brand: formality, tone, the brand's context and the words you never use. It is defined once, in writing, and applied the same way to every description, whether a person or an AI writes it.
Why voice gets lost
In a small brand, one person writes every product page and the voice comes naturally. As the catalog grows, several people write, suppliers send their own copy and new tools arrive. Soon one page addresses the buyer casually and another formally; one promises “incredible results” and another is a soulless technical list. Buyers don't know these are different people: they just feel the brand is not reliable.
What defines a brand voice
For a catalog, a useful brand voice fits in five decisions:
- Formality. Formal, informal or neutral, applied the same way on every page.
- Tone. How the brand should sound, in one or two sentences: “clear, practical and never exaggerated”.
- Context. What an in-house writer would know and a new one would not: what the brand sells, to whom, what sets it apart.
- Banned words. Terms that must never appear, for style or legal reasons: “miraculous”, “cure”, “the best”.
- Languages. Which ones you write in and which is primary, the one translations start from.
Written this way, the rules work for a person and for an AI alike.
Voice in a product record
A product page has little room for personality, and that is fine: voice shows more in what you avoid than in what you add. Some signs of a well-applied voice:
- The same words for the same things. If the brand says “cordless”, it does not say “wireless” on the next page.
- Measured promises. What is said can be backed up.
- The same order of information. Buyers find the data where they expect it.
- The same level of detail for similar products.
Variants and consistency
Variants are where a missing voice shows most: nine sizes of the same product with nine different descriptions. The right approach is for the description to live on the parent product, or for each variant to start from it, so they all say the same thing.
Test before generating everything
Before applying the voice to thousands of records, test it with a few real products and read the result. Adjusting a rule with five examples is far cheaper than fixing five thousand descriptions later. Read the results out loud if you can: a sentence that sounds wrong when spoken usually reads wrong too, and it is the fastest way to spot a voice that drifted.
Common mistakes
- A voice that only lives in someone's head. If it is not written down, every new writer reinvents it.
- Contradictory rules. “Friendly but formal” leaves everyone to interpret it differently; pick one.
- Adjectives instead of rules. “Inspiring” cannot be applied; “short sentences and no exclamation marks” can.
- Forgetting banned words. They are the easiest part to check and the one that prevents the most trouble, especially in industries with rules about what can be promised.
- Never updating it. When the brand changes, the catalog's voice should change too.
How Bruxel handles it
In Bruxel the brand voice is configured once and applies to the whole catalog: output languages and the primary one, formality, tone, the brand's name and context, and banned words. The template decides which fields are generated and the voice decides how each one is written, and a preview with real products lets you see the result without spending credits. The brand voice guide explains it, and the generating descriptions guide shows the next step. To go deeper on the copy, read how to write product descriptions that sell.
Where to start
Gather ten product pages you like and ten you don't. Write down what the good ones have that the others don't: it is almost always the formality, the tone and two or three words the brand never uses. Those notes are your first brand voice.
The video tutorial
Chapters
Video transcript
In this video, you'll tell Bruxel how your brand writes.
You set your voice once, and it applies to your whole catalog.
The template decides which fields get generated; your voice decides how each one is written.
First, which languages you want your product records in.
Then the form of address, which matters in Spanish: informal, formal or neutral.
The tone says how your brand sounds.
And the context: what someone on your team would know and the AI wouldn't.
Forbidden words never show up in anything that gets generated.
Save, and the voice applies from the next record you generate.
And by default, Bruxel writes without medical claims or absolute guarantees.
To hear it, generate a product. The button shows the cost before you're charged.
The record now speaks to your customer in your tone and explains how to use it, without promising results.
In the next video, you'll generate your first descriptions in a batch.
Frequently asked questions
What is the difference between voice and tone?
Voice is the brand's personality and does not change; tone adapts to the moment. A brand can have a friendly voice and use a more serious tone in safety warnings.
Do I need a long style guide?
No. For a catalog, a few clear rules are enough: how to address the reader, how it should sound, what to know about the brand and which words are banned. Nobody consults a fifty-page guide.
How do I check that the voice holds?
With a sample: take ten random product records every week and read them as if you were the buyer. If one sounds like another brand, the missing rule is usually obvious.
Sources
About the author
Manuel Mattar
Founder of Bruxel.AI
In 2014 he founded Wolf Sellers, an e-commerce agency in Mexico City, and later Bruxel.AI, the AI-powered PIM.
See all posts by Manuel MattarTry it with your catalog
Create your free account and try Bruxel with your own products using the welcome credits.
Keep reading
· 3 min read
AI in your product catalog: where it helps and where you need human review
AI already writes descriptions, fills attributes, proposes categories and translates. Where it pays off, where it fails and how to keep human review.
· 3 min read
AI product photos: what works today and what doesn't
AI can already put your product in a scene or in a model's hands, but it does not replace the real photo. What works, what fails and how to disclose it.
· 3 min read
Alt text in e-commerce: why it matters and how to write it
Alt text makes your product images useful to people who cannot see them and to search engines. What to say, what to avoid and how to write it at scale.