Chargeback and Showback

Central software budgets produce no incentive to economise. Attributing cost to consumers does, and it introduces its own distortions.

The two mechanisms

Showback. Costs are attributed and reported to teams. Nobody's budget changes. Informational.

Chargeback. Costs are actually charged to team budgets. Consequential.

Showback is the sensible starting point and frequently sufficient. Chargeback changes behaviour more and creates the incentives described below.

What attribution requires

Cost per application, which you have from the spend work.

Users per application, by team, which requires your identity data joined to your directory.

An allocation rule for shared costs — platform fees, applications everyone uses, the management overhead itself.

A cadence, monthly or quarterly.

The join is the work. Application to user to team to cost centre, maintained as people move.

What showback produces

Teams discover what they cost. The reaction to a first showback report is reliably surprise.

Unused seats get returned, because now there is a reason to.

Duplicate purchases become visible to the people making them.

Requests get more considered, because the requester sees the number.

Little of this requires enforcement. Visibility does most of the work, which is the argument for starting there.

What chargeback adds, and costs

Adds: real budget pressure, and a genuine decision about whether a tool is worth its cost to the team that uses it.

Costs: disputes about allocation, gaming, and a category of behaviour where teams avoid tools that would help them because the cost lands on them and the benefit lands elsewhere.

The specific failure: a team declines to adopt a security or compliance tool because it is charged to them. Anything the organisation wants universally adopted should be funded centrally, not charged.

Allocation questions that cause arguments

Shared platforms everyone uses. Per head is simplest and it penalises large teams for existing.

Applications used by one team on behalf of others.

Users in more than one team.

Contractors and part-time staff.

Committed capacity nobody uses. Charging teams for seats bought in a block that they did not request is the fastest way to discredit the whole scheme.

Decide the rules once, publish them, and change them rarely. Most disputes are about rules being applied inconsistently rather than about the rules themselves.

Making it useful rather than administrative

Report per team: total, per head, trend, and the top five applications.

Compare like teams, which is where the interesting variation shows.

Attach the unused-seat figure, which converts the report from a bill into an action.

Give teams a route to act. A report with no mechanism to return seats produces resentment rather than savings.

Keep it to one page per team.

The first showback report

The format that produces action rather than a complaint about the allocation method.

One page per team.

Total annual software cost, and cost per head.

Comparison to the median across comparable teams, which is what makes the number meaningful.

Top five applications by cost.

Unused seats and their value, which is the actionable part.

A single sentence on how to act: who to contact to return seats.

Send it to the team lead first, before any wider distribution. A leader who sees their number in a group setting for the first time defends it; one who saw it privately a week earlier has usually already acted.

What to fund centrally regardless

Charging for some things produces exactly the behaviour you were trying to prevent.

Security tooling. A team declining an endpoint agent because it lands on their budget is a worse outcome than the cost.

The identity platform, since federation benefits everyone and charging discourages it.

Compliance systems the organisation is obliged to run.

Anything mandated, where the team has no real choice and the charge is therefore a tax rather than a signal.

The management overhead itself, including the register and the person maintaining it.

Charge for discretionary tools where the team has a genuine decision to make, and fund everything else. The test is whether a team declining it would be an acceptable outcome.