AI Usage

Monitor token consumption, request volume, rate limits, and plan allowances for AI features.

AI Usage shows how much of your plan’s AI capacity you have consumed and how activity trends over time.

Overview cards

Typical metrics include:

MetricMeaning
Tokens usedTotal model tokens consumed in the billing period
RequestsNumber of AI API calls (assistant, banner, compliance, translations, etc.)
RemainingAllowance left before throttling or upgrade needed
Blocked requestsCalls rejected due to limits or policy
Success rateShare of requests that completed without error

Plans may also show daily caps separate from monthly totals.

Charts

  • Usage over time — spot spikes after large classification jobs or translation batches
  • Hourly prompt/completion split — see whether input or output tokens drive cost

Use trends to schedule heavy work (bulk translations, full-site compliance runs) outside peak marketing periods.

Feature breakdown

When available, a breakdown lists which product areas consumed tokens—for example Assistant chat, banner generation, or translation drafts. Prioritize tuning the highest-cost workflows first.

BYOK (bring your own key)

If your workspace uses your API keys (configured in AI Settings):

  • Some usage may appear under BYOK counters
  • Billing for those calls goes to your provider account
  • CookieShift may still meter requests for fair-use limits

Limits and upgrades

When you approach limits:

  1. Review AI Usage alerts or banners
  2. Open Billing to compare plans
  3. Reduce optional jobs (repeated compliance checks, large translation batches)
  4. Enable BYOK for high-volume teams comfortable managing provider keys

Good hygiene

  • Run one scan before batch AI classification—not one scan per button click
  • Approve translation drafts before requesting another locale
  • Cache human-reviewed banner copy in Consent Setup instead of regenerating daily