Playbook / Guide 9 min read

Running Theme Ops Across 25 Client Stores

One store is a project. Twenty-five is an operation — and the things that break are almost never the code.

TS
ThemeSync Team
ThemeSync with a store rail of client stores and the SŌMA store home
Stores in the rail, one store’s themes on the page.

What actually changes with scale

The work per store doesn't get harder. What changes is that context stops being free.

On one store you know what's live, what's in flight and what the client is waiting on, because it's the only thing you're doing. At twenty-five, none of that is in anyone's head. It has to be somewhere you can look, or it doesn't exist.

The failures that follow are consistent across agencies:

  • Credential sprawl. Theme Access passwords in a spreadsheet, a password manager, three .env files and a Slack thread from 2024.
  • Silent sync failures. A store's tracking broke six weeks ago. Nobody noticed because nobody was looking at that store.
  • Lost review rounds. A client was sent a preview in August and never replied. Nobody chased, because nobody owned the list.
  • Unverifiable offboarding. A developer left in June. Which client stores can they still write to? Nobody can answer.
  • Inconsistent process. Each store was set up by whoever was free, so each one works slightly differently.
The commercial risk is concentrated in the last two. "We can't tell you who has access to your store" is a genuinely damaging thing to say to a client, and on any store with a compliance requirement it can end the relationship.

Credentials: the thing to fix first

Every Theme Access password can overwrite a live storefront. Treat them like production deploy keys, because that's what they are.

Three properties that matter

  1. One credential per store. Never reuse across clients. A leak stays contained.
  2. Held centrally and encrypted. Not on laptops, not in chat, not in a shared sheet. Central storage is what makes an audit possible at all.
  3. Revocation is a checklist item. At project close, revoke and tell the client you did. Clients remember this — it's a differentiator worth almost nothing to deliver.
Where credentials liveCan you audit access?Can you offboard cleanly?
Developer laptops, local configNoNo
Shared spreadsheetSort of — who copied it?No
Chat messagesNoNo
Team password managerPartiallyPartially
Central, encrypted, per storeYesYes
The test to run today: pick a developer who left in the last year and list every client store they can still write to. If you can't, that's the first thing to fix — ahead of any tooling improvement.

Naming conventions do more work than they should

Unglamorous and disproportionately effective. At scale, a name is often the only context available to whoever finds a thing.

Names that answer questions without asking anyone
# Themes on a client store
LIVE — Dawn 15.2
STAGING — checkout revamp — 2026-10
BACKUP — pre-BFCM — 2025-11-14 — delete after 2026-02
AGENCY — homepage refresh — delete on launch

# Branches
production                      # mirrors the live theme
feature/plp-filters
release/2026-10-12

Three questions a good name answers: what is this, when was it made, and when can it go. The third is the one everyone omits and the one that makes cleanup possible by someone who wasn't there.

Apply the same discipline to how work is labelled. If a theme is called "homepage refresh" everywhere — branch, theme, review link, invoice line — then anyone can follow the thread without asking.

Sync health has to be a dashboard, not a memory

Tracking a live theme into git is only valuable while it's working. A sync that silently stopped is worse than no sync, because you'll trust a record that stopped being true.

Four states worth distinguishing:

StateMeansAction
HealthyPulled recently, no errorsNothing
SyncingRunning nowNothing
StaleHasn't pulled in longer than expectedCheck credentials and interval
ErrorLast attempt failedFix before any deploy to that store

The requirement is seeing this across all stores at once. Checking twenty-five stores individually is a task nobody does, which means the information may as well not exist.

Treat "error" as a deploy blocker. If tracking is broken you don't know what's on the store, which means you can't know what your deploy will overwrite. Fix the sync first.

Roles, so access doesn’t depend on who’s free

Small teams share one login and it works fine until it doesn't. Roles matter for two practical reasons, neither of which is bureaucratic.

ThemeSync team page with members and roles
Roles defined once, not per person per store.

Onboarding and offboarding become events, not projects. A designer joins and gets designer access to the stores they're on. They leave and it's removed. Without roles, both operations are archaeology.

Sign-off means something. If everyone has the same access, "approved by QA" is just a person's name. If QA is a role, an approval carries information about who was competent to give it.

Standardise these four and scale stops hurting
  1. Credentialsauditable

    One per store, central, encrypted, revoked at close.

  2. Namingself-documenting

    What it is, when it was made, when it can be deleted.

  3. Sync healthone view

    Healthy, syncing, stale or error — across every store on one screen.

  4. Rolesclean offboarding

    Design, QA and PM access defined once, not per person per store.

Boring, cheap, and it removes most of what breaks at 25 stores.

Client communication at portfolio scale

The operational failure clients actually notice isn't a broken sync. It's silence.

A preview goes out, the client doesn't reply, and it drops off everyone's radar because no one owns the list of outstanding reviews. Three weeks later the client asks why their project hasn't moved — having been the blocker — and it still reads as your fault, because you didn't chase.

Two things fix most of this:

  • A single view of what's awaiting a client. Which reviews are open, which pages are unapproved, how long they've been waiting. That's your chase list and it should take one glance.
  • Stable review links per store. If every round needs a fresh link, every round has a moment where the client can't find the current one — and that's where projects stall.
What clients experience as competence is small and consistent: the link always works, you know what's outstanding, and you tell them when access was revoked. None of it is technically hard. All of it requires the information to be somewhere you can see.

How ThemeSync supports a portfolio

  • Credentials are stored per store, encrypted at rest. Access is centralised rather than distributed across laptops and chat threads, so an audit is possible and offboarding is verifiable.
  • Stores are first-class. A store switcher plus a per-store home means "what's happening on this client" is a place you go, not a thing you reconstruct.
  • Sync state is visible per store. Healthy, syncing, stale or error, so a broken tracker surfaces before it becomes a bad deploy.
  • Untracked live themes are flagged. Stores whose live theme nobody is watching get surfaced rather than quietly forgotten.
  • Team members have roles. Designer, QA and PM access is defined once, and approvals record which role signed off.
  • Feedback is summarised per theme. Open comments and outstanding approvals are visible, so the chase list is a view rather than a memory.
  • Review links are stable per theme. The client keeps one URL across revision rounds.

Takeaways

  • Scale doesn't make the work harder — it makes context expensive. What isn't visible somewhere stops existing.
  • Fix credentials first: one per store, central and encrypted, revoked at project close.
  • If you can't list what a departed developer can still write to, that's the most urgent gap you have.
  • Naming conventions are cheap and do a disproportionate amount of work. Always include when a thing can be deleted.
  • Sync health must be visible across all stores at once, and an error state should block deploys.
  • Roles make onboarding and offboarding events rather than investigations, and make sign-off meaningful.
  • Own the list of what's awaiting a client. Silence is the failure clients actually notice.

Twenty-five stores, one place to look

ThemeSync keeps per-store credentials encrypted, shows sync health across your portfolio, flags untracked live themes, and gives every theme a stable client review link.

Try ThemeSync Free →