AI features in Elovate — such as translations, content generation, and enrichment — use “tokens” from your connected AI provider (e.g., ChatGPT/OpenAI). Understanding how tokens are counted helps you manage costs and avoid unexpected interruptions.
1. What are AI tokens?
A token is a unit of text used by your AI provider to process input and generate output.
Roughly, 1 token ≈ 4 characters of English text (including spaces and punctuation).
Both your prompt (input) and the generated text (output) count toward your total token usage.
Example:
If you send 500 tokens of source text and the AI generates 1,000 tokens in output, the total cost is 1,500 tokens.
2. How token usage works in Elovate
Each AI job (translation, enrichment, etc.) sends your selected attributes and prompts to the provider.
Longer product descriptions, more attributes, or complex prompts = higher token use.
Bulk or continuous jobs may process thousands of products at once, so usage can add up quickly.
3. What happens if you run out of tokens
If your AI provider account runs out of tokens (or credit balance), Elovate will:
Stop processing the AI tasks mid-job.
Show errors in the job overview for affected items.
Log the failure in Apps → Synchronisation logs, where you can click the entity to troubleshoot.
You’ll see a notification in Elovate for GPT-related token errors.
4. How to fix token errors
Add credits to your AI provider account (e.g., OpenAI billing page).
Retry failed items:
From the job view, filter for errors and click Regenerate.
From the item view, open it, adjust if needed, and click Regenerate translation/content.
5. Best practices to reduce token usage
Keep prompts concise.
Only send attributes that need AI processing.
Split very large jobs into smaller batches.
For translations, exclude attributes that don’t need to be translated (e.g., SKU, internal notes).