Articles in this category

GOOD STUFF!

Optimizing for Performance and Credit Usage

Andrija Krstic Updated by Andrija Krstic

This article summarizes the best tactics on how you can optimize your work with an AI agent to get the best output while keeping your credit usage low.

Before you start

Familiarize yourself with terminology – Before you begin prompting, review the Glossary of Terms. This will help familiarize yourself with the terminology used and will help you understand some of the most common terminology behind an AI.

Plan your site’s structure upfront – Sketch or outline your page sections before starting the build. Decide on the number of pages, their hierarchy, and what each section should contain. A planned structure minimizes exploratory prompting and allows the agent to execute your vision efficiently. Head over to Webpage Anatomy to understand all of this better.

Write lean Project Instructions – Treat these as the agent’s long-term memory. Keep only essential brand, typography, and layout rules (under 2,000 characters) so you don’t waste tokens re-teaching context in each chat.

Pick the right AI model – Use Core for simple tweaks, Advanced for most styling and copy updates, Pro for multi-section or data-driven edits, and Expert only when the problem is complex enough that it’s cheaper to solve in one pass.

Use built-in tools to optimize the spend

Use the Element Selector and Visual Editor – Perfect for changing text, colors, padding, or margins directly in the interface. These actions don’t burn credits.

Leverage the Prompt Enhancer – It reformats and clarifies prompts automatically before you send them. Using it is free and reduces retries.

Reference resources instead of describing them – Use the “@” handle to link products, bundles, or surveys instead of pasting details manually.

Upload screenshots for inspiration and clarity – When you have an example of a desired section, upload or paste a screenshot directly into chat. Agent can visually interpret it and reconstruct a similar layout faster than deciphering long descriptive text.

Connect checkouts efficiently – Use Element Selector to pick each CTA and the “@” resource handle to bind the correct product checkout in one pass.

Prompting tactics that cut spend

Plan scoped chats – Start a new chat for each page or section. Shorter contexts consume fewer tokens and stay easier to manage.

Be explicit and concise – Give narrow instructions like “Change the hero headline text only; leave spacing untouched.”

Group related edits intelligently – Combine small, related changes into one precise prompt rather than many fragmented ones.

Avoid conversational or emotional phrasing – Stick to factual, directive language so the agent interprets your intent efficiently.

Keep prompts deterministic – Note what must not change (“Keep all existing IDs and class names”) to prevent collateral edits that require correction prompts.

Versioning and recovery

Use Version History to roll back – Restore an earlier version from the chat’s history instead of paying to “undo” with new prompts.

Delete a bad prompt to revert its entire change set – This instantly resets the site to the pre-prompt state (and removes later dependent changes) without spending credits.

Fork when exploring alternatives – Spin a branch from a known-good version to compare approaches side-by-side without burning credits on “rebuild it like before.”

Execution hygiene and extras

Prefer small, testable deltas – Ask for one section or component at a time, review, then proceed if all OK. This minimizes token-heavy retries.

Use the admin’s “Your usage” link – Track remaining credits in real time on My Plans and adjust model/prompts accordingly.

Reserve Expert for thorny tasks – Major layout overhauls, complex animations, or multi-file restructures. It’s slower and pricier but may still net cheaper if it solves the problem in one go.

Do pure styling touch-ups in the visual editor first – If you can achieve the look with CSS tweaks, do that credit-free before prompting for structural changes.

Annotate what must not change – “Apply these changes only to mobile version of the site.” Avoids accidental cascade changes that trigger more prompts to fix.

Understand what consumes credits – Only executed prompts cost tokens. Visual Editor and Prompt Enhancer are free.

Use the Add-on flexibly – AI Add-on can be paused when you’re not building. Reactivate it during active development periods to control spend.

Did you find this resource helpful?
Return to top
Ready to get started
with Subbly?
Try for free