The Backend Audit Playbook

Back-office work for a service business is mostly movement between systems: a lead arriving in one place, an invoice living in another, a report assembled from four. The hours are not in the work itself — they are in the seams between the tools.

• Data handoffs — the same information copied or re-keyed from one system into the next.
• Reporting lag — numbers assembled by hand on a schedule, so decisions run on stale data.
• Routing decisions — someone deciding what happens next, and who does it, by hand.
• Duplicate entry — the same fact typed twice because two tools don't talk.
• Follow-up — chasing leads, invoices, or tasks manually, with nothing reminding anyone.

Every engagement at Crest starts with a map, and yours should too. Pick one representative week and write down every step of your operation: what starts it, who touches it, which tool it lives in, and how long it takes.

Use the Backend Map CSV below — one row per step. Be honest about the spreadsheet between the tools; it is usually doing more work than anyone has admitted.

• A step is anything a person does more than once a week.
• Owners matter more than order. If a step has no owner, that is a finding, not a detail.
• If a step takes under ten minutes but happens daily, count the weekly total. Small steps are where the hours hide.

List every tool you pay for: what it is supposed to do, what it actually does, what it costs per year, and how many human steps sit between it and the next tool. Then look at the seams you mapped in chapter 2 — the value of a tool is measured in the seams it removes, not the features it lists.

A tool that does its job and touches nothing else is fine to rent. A tool that requires a human to connect it to the rest of your operation is a handoff wearing a subscription.

An owned backend is one system that runs the operation: leads routed to the right next action, delivery tracked against scope, reporting assembled live from your data, follow-up that happens because the system decides it should. You own the code, and you can change it without waiting on a vendor queue.

The decision rule we use with clients: keep renting commodity tools that do their job; build the parts that are actually your operation. The renewal question in chapter 5 is where most owners discover which is which.

Renewals are the moment the decision gets made by default. Before you sign anything, answer these eight in writing. If a vendor cannot answer one of them in writing, that is the answer.

• What is the renewal date and the notice date — and who holds the reminder?
• What does the contract say about price increases, at renewal and after?
• Which seats or tiers did we actually use in the last 90 days?
• What data does the tool hold, and how do we export all of it?
• Which steps in our map does this tool remove — and which ones still need a human?
• What would it cost to own the equivalent instead of renting it?
• How long would switching take, and what breaks in the meantime?
• If we do nothing, what renews at what price — and what does that buy us?

The report assembled every Friday from four exports is not a report — it is the visible cost of the seams. Add up the hours your team spends collecting, cleaning, and re-keying numbers for it. Most owners we audit find the reporting tax is larger than one full-time hire, once the seams are counted.

The test is simple: if the numbers already live in your systems, they should assemble themselves. A backend that reports from your own data removes the tax without adding a subscription.

You do not need a project to start. You need a calendar with four blocks.

• Weeks 1–2: map. Complete the Backend Map. Name the top five seams.
• Weeks 3–4: cut. Cancel unused seats and tools. Every cut simplifies the map before you automate anything.
• Weeks 5–8: automate the first two seams, smallest first. One working automation beats five planned ones.
• Weeks 9–13: measure. Re-run the map. The hours are the scoreboard.

This playbook is the free audit, abbreviated. The full Crest audit does it for you: we map your backend, score every step, name the automation candidates in order of return, and put the plan in writing — the map, the build, and the fixed price.

If the audit does not find a backend worth owning, you owe nothing and keep the map. That is the same guarantee we give every client, in writing, before any build.