How much custom development budget should a $20M-$500M merchant expect to set aside annually just for ongoing theme customization, separate from the initial build?
Annual theme customization runs 15 to 20% of the original build cost in upkeep alone, roughly $4,000 to $12,000 against a $25,000 to $60,000 section library (Deploi estimate, illustrative, August 2026). New campaign and roadmap work sits on top of that, and for a $20 million to $500 million merchant with real campaign tempo it is the larger of the two lines.
Two budgets, not one, and conflating them is why this number always looks wrong
Teams ask for "the theme budget" and get a single figure that either underfunds the roadmap or overfunds maintenance. Split it.
| Line | Band | What it covers |
|---|---|---|
| Upkeep | 15 to 20% of build cost per year, roughly $4,000 to $12,000 on a $25,000 to $60,000 library | Theme version merges, regression fixes after app installs, browser and device drift, accessibility and performance regressions, small copy and setting changes |
| New work | Sized by campaign tempo, not by build cost | Net-new sections, seasonal art direction, PDP and collection experiments, new template types |
| Standing capacity | A named lane, not a pool of hours | The thing that keeps the queue from becoming a quarterly emergency |
Bands are Deploi estimates, illustrative, August 2026. They are planning figures from a small sample of engagements, not published rates.
What drives the upkeep line up
- App count. Every install is a potential regression on a template you did not touch. App widgets break at theme updates and at odd breakpoints, and the fix queue often runs through vendor support (community-reported theme, Deploi Build vs. Buy analysis, August 2026).
- How customized the theme is. Code edits either carry across a theme update or they do not, and when they do not you copy them to the new version manually (per Shopify's Help Center, September 2026). A heavily edited premium theme turns every vendor release into paid merge work.
- Whether you take updates at all. Skipping them is free this year and expensive in year three.
- Locales and markets. Each one multiplies the QA surface.
What drives the new-work line
Campaign tempo and merchandising ambition, and nothing else. A brand running one campaign a quarter and a brand running one every two weeks can have identical builds and a fourfold difference in this line. Size it from the marketing calendar, then work backward, rather than from a percentage of anything.
Honest boundaries
- These are planning bands, not quotes. Scope, agency, market and locale count all move them.
- They exclude replatforms, redesigns, new sales channels, and anything with "headless" in the title.
- They assume a section library that marketing can actually use. If every campaign page needs a developer, the new-work line is not a budget, it is a bottleneck with a dollar sign in front of it, and the fix is upstream on the parent page.
When this budget is the wrong instrument
When the requests are all variations of five things. Funding an annual customization budget to absorb repeat variation is paying rent on a problem that a parameterized section retires. Spend the first year's new-work budget on the library instead, then set upkeep and a smaller new-work line against it.
The Deploi point of view
Our own position, from building on Shopify. Separate from the facts above.
- Our take: Fund a small standing lane rather than a project budget you release quarterly. New section requests land in a dev queue, and without a named lane self-serve velocity quietly dies, which is the failure mode that sends marketing back to a page-builder trial.
- What we’ve seen: The teams who skip the upkeep line do not save the money. They spend it in year two, in larger and less convenient pieces, usually as an emergency fix to a storefront that drifted three theme versions behind while nobody was looking at it.
- What it takes: roughly 200 hours for an art-directed hero and banner system and roughly 150 hours for collection merchandising and product rails, if a system is added in-year (directional Deploi estimates from a small sample of engagements, not measured averages).
- Where we disagree: The common advice is to budget theme work as a percentage of revenue. That produces absurd numbers in both directions at this band. Budget upkeep as a percentage of build cost, and budget new work from the marketing calendar. They are different questions with different drivers.
- What this page adds: the split between upkeep and new work, the percentage that sizes the first one, and the specific drivers that move each.
Reviewed by Martin Dejnicki, Director of SEO & AI Search. Facts verified 2026-09-13.