Agencies

Agencies do not have a time problem; they have an attribution problem. The hour happened. The question is which client pays for it, at what rate, against which phase, and whether anybody will argue about it at invoicing.

Which means the tracking that matters is the tagging, and a system that captures duration accurately and attribution loosely has solved the easy half. For related guidance and industry context, see GitHub.

Four places, in order of size

Unbilled scope. Work done, agreed informally in a call, never added to the estimate. It appears as a utilisation problem and it is a contracting problem.

Retainer overrun found late. A retainer burned through in week three of a month nobody was watching, discovered at invoicing when the conversation is hardest.

Non-billable drift. Internal work expanding without anybody deciding it should.

Rounding and write-offs. Individually trivial, collectively a percentage point or two of margin.

Only the fourth is a rounding question. The first three are visible in a time record and invisible in an invoice.

• Retainer burn against the month, per client.
• Hours logged with no client attached.
• Billable proportion by person, as a trend rather than a target.
• Write-offs, with a reason recorded.

The third comes with a warning. A billable percentage that becomes a target stops being a measurement: people attribute generously, and the number improves while the margin does not.

Most agencies price from a blended rate and measure against it. The useful comparison is per client and per phase, because the blended figure conceals the account that has been unprofitable for eight months.

That analysis needs one thing the timesheet must supply: the attribution has to be right at the point of entry. Reallocating hours at month end to make a report balance produces a report that balances.

Agencies run on contractors, and contractor hours arrive in whatever format the contractor uses. Bringing them into the same attribution is worth more than any internal reporting feature, because unattributed contractor cost is where the account that looked profitable turns out not to be.

It also raises a question worth settling with an adviser: the degree of control an engager exercises over how somebody works can bear on their employment status, and requirements about how time is recorded are a form of control. Which is a question for your own counsel rather than for this page.

• Can a person see everything recorded about them?
• Is there any collection they are not told about?
• Can an administrator alter a record, and is the alteration logged?
• Can we export everything if we leave, in a form that is readable without you?

The third is the one nobody asks. A time record an administrator can silently change is not evidence, which matters on the day a client disputes an invoice.

Most agencies round to some increment. Whatever yours is, it should be stated in the client agreement and applied consistently in both directions.

Rounding applied only upward is the kind of thing a client discovers once and remembers, and it is discoverable from an itemised record of the sort this software produces. Which is an argument for deciding the policy before the data exists to examine.

Some agencies share time detail with clients and some do not. Both are defensible; what is not is sharing a version assembled for the purpose.

If a client will see the record, the people entering it should know that from the start, because it changes how they write a description and it is not a thing to discover afterwards.