We get asked the headless question a lot: should we go headless?
The honest answer is: it depends, and most of the time the question is phrased too early. Headless is a useful tool when your storefront's constraints really are limiting the business. It's expensive complexity when they aren't.
When headless is the right call
You'll get real value from a headless rebuild when:
- The frontend is slow in measurable ways that hurt conversion, and the monolith's templating layer can't fix it.
- Your team wants to ship multiple distinct experiences (regions, brands, apps, kiosks) and the monolith is fighting you.
- You have engineers who want to own the frontend. Headless trades out-of-the-box capability for control.
- Speed-to-launch isn't the bottleneck — you have 4–9 months of runway.
When it isn't
You'll regret going headless when:
- Your storefront is fine and your conversion problems are upstream (price, product, traffic mix, checkout friction inside the platform).
- Your team is small and you don't have frontend ownership.
- You'd lose access to platform features (search, recommendations, promotions) you don't have time to rebuild.
- Your timeline is "next quarter."
A useful framing
We sometimes ask clients to fill in this sentence:
"If we don't go headless, in 12 months we won't be able to ____."
If the blank is hard to fill in honestly, headless probably isn't yet worth the cost. If the blank fills itself — different brand experiences, sub-second product detail pages, real-time inventory across channels — then the rebuild is doing real work.
The middle ground
A truly headless rebuild isn't the only way to modernize a storefront. We've also helped clients:
- Partially decouple by moving just the product detail and category pages to a modern stack while leaving the rest of the platform alone.
- Edge-cache aggressively to get most of the perf win without the rebuild.
- Adopt a design system that makes the existing storefront easier to change.
Each of these can buy you 12–24 months of runway and let the headless decision wait until it's obvious.