Media Planning and Buying Software: Closing the Month When Plan, Delivery and Invoice Are Three Different Numbers

Why media operations breaks at month end and nowhere else

It is the fourth working day of the month. A broadcast invoice has arrived for a client campaign that ran in three markets. The station billed 412 spots. Your affidavit shows 407, two of which ran outside the daypart the insertion order specified. The ad server says the digital half of the same campaign delivered 3 percent under the guaranteed impressions, so a makegood was agreed by email between your buyer and the publisher's account manager, and that email is in a thread nobody else can find. Your finance team needs to invoice the client in six days with commission calculated on a structure that is specific to this client's contract, and your accounts payable run to the vendors is due before you get paid.

Three numbers exist for every line: what was planned, what was delivered, what was invoiced. They never agree, and they are not supposed to. The job of a media operations system is to hold all three against one line and route the difference to a human when it exceeds a tolerance you defined. Almost nothing you can buy does exactly that in the shape your agency needs.

The stack is usually Excel or Google Sheets for the plan, email for insertion orders, Bionic Advertising Systems or a similar planning tool for some of it, Basis Technologies for programmatic execution, Campaign Manager 360 and platform native reporting for delivery, Mediaocean where the agency is large enough to carry it, and QuickBooks or NetSuite at the end. Each of those is real software doing a real job. What is missing is the spine that carries a line item from plan through insertion order through delivery through vendor invoice through client bill without anybody retyping it.

Problem 1: the insertion order is an email attachment, so no version is authoritative

A plan gets approved. The insertion order goes out as a PDF. The client shifts budget between two markets. A revision goes out. The publisher confirms verbally. Two weeks later a cancellation deadline passes on a broadcast buy and nobody is certain which version the station is holding you to. When the invoice arrives against version two while your finance team is billing from version four, the difference becomes an argument you have no documentation to win.

What a custom build does: the insertion order is an object with immutable versions, an approval chain, a status, and a diff between versions that anyone can read. Cancellation windows are stored as dates on the line with an alert before they pass, because broadcast and out of home carry real cancellation deadlines and missing one converts a client budget shift into agency liability. Publishers get a portal link or a generated document, and every version they were sent is recoverable in seconds. This alone settles most of the disputes that currently consume your operations lead.

Problem 2: delivery data arrives in as many shapes as you have vendors

Programmatic delivery comes clean from an API. Direct digital comes from Campaign Manager 360 or a publisher dashboard with a different impression counting methodology. Broadcast comes as an affidavit, sometimes as a PDF, sometimes as a printout scanned by an assistant. Print comes as a tearsheet. Out of home comes as a proof of posting photo and a play count. Podcast comes as a host read confirmation and a download report on a 30 day lag.

Off the shelf systems that handle this well tend to handle it well for one market's conventions. That is the fair criticism of the packaged category: Mediaocean is genuinely strong, and it is built around a specific set of buying and billing conventions, and it is typically priced against media volume, which means your software cost scales with a client win rather than with your team. If your channel mix or your market's conventions sit outside that model, you spend your life in workarounds and still pay on volume.

What a custom build does: an adapter per delivery source that normalises into one delivery record keyed to the insertion order line, with the source and the counting basis preserved so a discrepancy can be explained rather than just flagged. This is where AI does a genuinely useful job: broadcast affidavits and print invoices arrive as PDFs in a hundred layouts, and a document extraction pass turns them into structured spot lines with a confidence score, then auto matches to insertion order lines. In our builds the no touch match rate settles high enough that a media ops team stops keying invoices and starts reviewing exceptions, which is a different job with a different headcount.

Problem 3: makegoods are negotiated in email and never make it into the numbers

An under delivery gets settled with bonus weight next month, or a credit, or an upgraded position. That agreement changes what the client should be billed, what the vendor should be paid, and what next month's plan looks like. Because it lives in a thread, it usually reaches finance late or not at all, and the reconciliation for two months is wrong.

What a custom build does: makegoods are first class objects linked to the shortfall that caused them, with a type, a value, an approver, and a link to the compensating line in a future flight. Reconciliation then closes properly, because the shortfall is not an unexplained variance, it has a documented resolution attached. It also means you can answer the question a client procurement team eventually asks, which is how much under delivery occurred across the year and how it was made whole.

Problem 4: commission and billing rules are per client, and packaged systems assume otherwise

One client is billed gross with commission included. Another is net plus a fee. A third has a performance element. A fourth requires rebate treatment disclosed in a specific way under their contract. Some are billed in advance, some in arrears, some by market with separate purchase orders per market. Sequential liability language in your terms determines whether you owe the vendor when the client has not paid, and finance needs to see that exposure before it becomes a cash problem.

What a custom build does: a billing rules engine per client contract rather than a global setting, with the calculation stored on the invoice line so an audit two years later reproduces the number exactly. Client audit rights are common in media contracts, so traceability from a client invoice line back to the vendor invoice, the delivery evidence and the insertion order version is not a nice to have, it is the thing that makes an audit a two day exercise instead of a two month one.

Problem 5: nobody can see the exposure until finance builds a report

Committed spend against approved budget, by client, by market, by month, with cancellation windows and payables timing. Most agencies assemble that quarterly from spreadsheets, which means the answer is always about a position you were in several weeks ago. When a client pauses a campaign, the question of what is cancellable and what is committed needs an answer in an hour, not a week.

What a custom build does: the commitment is derived from live insertion order versions, so exposure is a query rather than a project. Cash timing comes from the same data, which is what lets a finance director see the gap between when vendors are due and when clients pay, per client, before it turns into a facility conversation with the bank.

What this costs and how long it takes

Across the 2,000 plus projects Digital Heroes has delivered, here is the honest shape. A focused first release covering the plan, versioned insertion orders with approvals, delivery ingestion from your main sources, and three way reconciliation with tolerance rules runs $80,000 to $170,000 and ships in 14 to 20 weeks. A full platform adding makegood workflow, per client billing rules with commission, vendor payables, sequential liability and exposure reporting, plus accounting system sync, runs $200,000 to $500,000 phased over 8 to 14 months.

What drives cost up in media operations specifically: the number of channels, because broadcast, print and out of home each carry their own document formats and their own reconciliation semantics, and none of them behaves like digital. Multi market operation, since currency, tax treatment and buying conventions differ and cannot be handled with a settings flag. The number of client contracts with genuinely distinct commission structures. And integration depth with your accounting system, where NetSuite and QuickBooks Online are different problems with different effort.

What keeps cost down: one channel group and your five largest clients for release one. Digital plus one broadcast market covers most of the reconciliation pain and all of the learning.

Build versus buy, and when Mediaocean is the right answer

Buy, and we will say it against our own interest. If you are a holding company shop operating in a market whose conventions the packaged systems were built for, with standard commission structures and heavy trading volume, Mediaocean is the right answer and building a parallel system is an expensive way to arrive at the same place. If you are a small shop buying only programmatic through one platform, Basis Technologies plus a spreadsheet is genuinely enough and a build is a vanity project. If your planning pain is the plan itself rather than the money, Bionic Advertising Systems solves a narrower problem for far less than a build.

Build when two or more of these are true. You are an independent with commission and billing structures that no packaged system expresses without workarounds. You operate across markets whose conventions differ. Your channel mix includes significant broadcast, print or out of home alongside digital, so no single platform holds the whole picture. Month end reconciliation costs you more than a week of skilled people. Or your software cost is priced on media volume and every client win makes the licence bill worse, which turns growth into a margin problem.

How to choose a developer for media planning and billing software

Ask them to model the line item on a whiteboard before you sign. The right answer has a plan line, an insertion order line versioned separately, delivery records from multiple sources with counting bases preserved, an invoice line, and a reconciliation record that links them with a variance and a resolution. A developer who draws orders and invoices has built ecommerce and is about to learn media on your budget.

Ask how they handle a makegood that spans two billing periods. If they treat it as a credit note and nothing else, next month's plan will be wrong and your reconciliation will not close.

Ask what they have actually integrated. A broadcast affidavit PDF is a different problem from a Campaign Manager 360 report, which is different again from a NetSuite journal posting. Ask for the named source and the named document type rather than a general claim.

Ask who owns the code and get it in writing before kickoff, including the repository, the cloud accounts and any extraction models trained on your documents. At Digital Heroes the client owns it from the first commit, and in a business where your billing logic is your commercial model, anything less is an unacceptable dependency.