Catch scope creep before it never hits the invoice

Who this is for

This playbook is for agencies and professional-services firms: studios, consultancies, implementation shops, retainers with a statement of work, and any owner-operated company where delivery people can add work faster than finance can see it. If “we did extra” and “it is on the invoice” are different conversations, you are in the right place.

It is not a collections-agency playbook and it is not an accounting close checklist. If the invoice already went out and the customer is slow to pay, that is a different path. If the pain you feel is the kickoff chase after a signature, start at client onboarding. If you cannot even say who is booked this week, read team capacity first — overcommit and unbilled scope often travel together.

Symptoms

If several of these are true in the same month, this is a live operational leak — not a tooling preference.

• A delivery lead can name three extras from this month that never became a change order.
• Time, tickets, and the statement of work live in three tools, so nobody can see the gap in one sitting.
• Scope questions are answered in Slack and the thread is treated as approval.
• Invoices are assembled at month-end from memory, a dump of hours, and a hope that nothing was missed.
• A client disputes a line after the invoice is sent because they never saw the extra work named.
• Someone proposed auto-invoicing every logged hour, including items the account lead already knows are contested.
• Finance asks delivery “what should we bill?” and delivery asks finance “what did we already bill?” in the same week.

What done looks like

Done is a working operating change, not a purchased seat or a dashboard nobody opens.

• Every engagement has a written statement of work a delivery person can open without asking the seller.
• Delivered work is captured in the same week it happens — hours, tickets, or a short delivery note — not reconstructed at close.
• Out-of-scope items are flagged as such before anyone drafts an invoice line.
• A named person must approve an out-of-scope line in writing before it is billed.
• A weekly work-in-progress review exists: what was delivered, what is flagged, what is approved, what is ready to invoice.
• Disputed and out-of-scope items cannot enter an automatic invoice run.

The leak is before the invoice, not after it

People hear “billing reconciliation” and picture aging reports, reminder sequences, and a collections tone. Those jobs matter. They are not this leak. This leak is quieter: the work already happened, the client already has the artifact, and there is no line waiting to be sent. You cannot collect what you never invoiced.

Treat the gap as an operating path, not a personality problem. Delivery people are not “bad at billing.” They are inside the work. Finance people are not “slow.” They cannot see a Slack yes that never became a record. The chain has to make the extra work visible while it is still cheap to name.

Keep invented leakage percentages out of the room. You do not need a made-up “we lose X percent of revenue” slide to justify looking. If extras exist and they do not have a written home, the path is already leaking. Describe the extras. Do not decorate them with a number you did not measure.

Write the chain: capture, flag, approve, invoice

Four steps. Name them. Put an owner on each. If a step has no owner, it is a wish.

Capture

Capture is not a perfect timesheet culture. It is a same-week record that a second person could recognize: what was delivered, for whom, against which engagement. Hours, tickets, a delivery note, or a checked milestone can all work. The test is whether someone other than the doer can find the work next Tuesday.

If capture only happens when finance nags at month-end, you will always reconstruct. Reconstruction is where scope dies. People forget the extra revision. They remember the mood of the client and decide not to “nickel and dime.” That decision belongs after a flag, not instead of a record.

Flag scope

Flagging is a comparison, not a feeling. Open the statement of work. Ask: is this line inside what we sold? If yes, it is ordinary delivery. If no, mark it out of scope before anyone writes invoice language. If you cannot tell, that is also a flag — the statement of work is too vague to bill against.

Do not make the flagger also be the closer. A senior delivery person can mark “looks extra.” A commercial owner decides whether to write a change order, absorb it, or stop the work. Mixing those jobs is how extras become silent gifts.

Approve

Approval is a written yes from someone who can commit the client or the firm. Slack thumbs are not that. “They seemed fine with it on the call” is not that. The artifact can be short: a change-order line, an email the client replied to, a signed note in the system you already use. The point is that a third person could find it later.

Absorbing scope is a valid approval. Write “we will not bill this” with a name and a reason. Silence is not absorption. Silence is how the same extra comes back next month and still has no home.

Invoice

Invoice only what is captured, in-scope or approved, and not in dispute. That sounds obvious until someone connects the time tool to the billing tool and calls it done. An automatic run that cannot see a flag will bill the fight and the gift in the same batch.

Weekly WIP, not only month-end

Month-end is too late for scope. The client has already used the extra work. The team has already moved to the next sprint. The person who did the revision is on another account. A weekly work-in-progress review is the cheapest control you can add without buying software.

Keep the review short. For each live engagement: what was delivered this week, what is flagged, what is waiting on approval, what is ready to invoice, what is blocked. If the meeting cannot answer those five questions, the capture step is still a memory job.

Who sits in it depends on the firm. A producer and a bookkeeper can run it. An owner and an account lead can run it. Do not invite the whole studio. Do not turn it into a status theater. The output is a short list of flags and approvals, not a slide.

Capacity and billing belong in the same week. If the team is already overcommitted, extras are how the week “still fits.” Read know who has capacity before you overcommit the team when the WIP review keeps discovering work that was never planned. Saying no to unpaid extras is easier when you can see the calendar cost.

Join the three pictures without a new platform

Most firms already have a statement of work, a place hours or tickets live, and a place invoices go. The leak is that nobody sits those three next to each other. You do not need a fourth platform to start. You need one view a person can read.

A usable first view is boring: engagement name, sold scope in one sentence, this week’s captured work, open flags, last invoice date. A spreadsheet is allowed if it has an owner and a weekly refresh. A dashboard is not required. A dashboard that pulls three disagreeing systems will just argue faster.

If CRM, billing, and delivery already tell three stories about the same client, stop. That is a source-of-truth problem, not a billing-automation problem. Read stop running the business from spreadsheets and inboxes before you connect an auto-invoice. Automating a dual picture invoices the wrong story.

Onboarding belongs here too. If the statement of work never lands in a place delivery can find, every extra is an argument about what was sold. Automate client onboarding without the email chase is the path that puts sold scope where the team can see it before week two.

What people usually get wrong

They wait for month-end and call the reconstruction “reconciliation.” Reconciliation that starts from memory is storytelling. Move the look to the week the work happened.

They treat every extra as either always billable or never billable. Both are lazy. Some extras are gifts you choose. Some are change orders. Some are a sign the statement of work was sloppy. The flag exists so you can choose in the open.

They auto-invoice because the integration demo was clean. Integrations do not know about a disputed revision or a verbal “we will eat this.” If the run cannot skip a flagged line, it is not ready.

They buy a PSA or a new finance suite to avoid writing the chain. A new system with the same Slack-yes culture will produce prettier unbilled work. Write capture, flag, approve, invoice first. Buy software only when that chain is real and the remaining pain is volume.

DIY vs hire

DIY the chain on the next billing cycle. Pick three live engagements. Write the sold scope in one sentence each. Ask delivery what shipped this week. Flag anything that does not fit. Get a written absorb-or-bill decision before you draft invoices. That is a week of calendar time for an owner or operations lead who already sits in the work.

Hire help when finance and delivery cannot agree on what “scope” means, when several tools disagree on the same engagement, or when someone is about to turn on auto-invoice across the book. The $997 assessment is a live consultation and a written analysis. It can name two to four improvements you run yourself — a weekly WIP, a flag field, a no-auto-invoice rule — and a later roadmap quoted only if you want us on integrations or a billing build. Implementation is not in the fee.

If several leaks are loud at once — unbilled extras, a messy data picture, a team that is already overcommitted — start with rank AI work that pays before you buy more tools. Billing software is a purchase. Ranking is how you decide whether this is the first purchase.

Controls before automation

Recommendations that touch customers, money, contracts, private data, or systems of record need human review, limited permissions, a log, and a fallback. Do not automate a broken path because the tool is ready.

• Never auto-invoice a line marked disputed, out of scope, or waiting on approval.
• Do not treat a Slack reaction or a call recap as commercial approval.
• A person who can commit the firm must sign the absorb-or-bill decision.
• Time and ticket tools inherit the access the human already had — not a wider finance key “so the sync can finish.”
• Keep a fallback: if the billing tool is down, a person can still assemble an invoice from the weekly WIP.
• If delivery, CRM, and billing disagree on the engagement, stop the automatic run until one sentence is true.

How this sits next to the other playbooks

This is one room in the fourteen mid-market AI asks: make the business answerable about work and money. Capacity tells you whether the extras were even possible. Data consolidation tells you whether the three pictures can sit in one view. Onboarding puts the sold scope where delivery can find it. Ranking decides if billing is this month’s leak or a later build.

If you want that ranking done with you, on a live call, and written down so the team can run the first weekly WIP themselves, book the assessment. Bring one messy engagement. Leave with a chain, not a new finance platform.