Roofing Back Office

Administrative Support

Data Entry for roofing contractors

A wrong field costs a few seconds to fix while the document is still open, and grows more expensive every time somebody downstream believes it.

Nate Jones · Founder, Roofing Back Office

Nate Jones is the founder of Roofing Back Office. On this desk he works with contractors whose reports are wrong in ways that were introduced months earlier by somebody being helpful.

Last updated · Reviewed by Nate Jones

What the fields say afterwards

Fields that say what the document in front of them said. That sounds like a low bar and it is the entire job, because the ways of failing it are subtle: keying from a summary somebody made, carrying a value forward from the last similar record, or completing something the source left ambiguous because leaving it blank felt unfinished.

Alongside that, a visible queue of everything the source did not actually say. Anything unclear comes back as a question rather than going in as a best guess, which means the exceptions are collected in one place instead of being distributed invisibly through your records as confident-looking values nobody will ever suspect.

An error gets more expensive every time it is read

At the keyboard, a mistake is nearly free. The document is open, the person who typed it is looking at the source, and correcting it takes a moment. Nothing else has consumed it yet.

One step out, a job-costing report is slightly wrong and nobody knows. Two steps out, a margin looks acceptable on a job type that is actually losing money, and the company prices the next one against it. Three steps out, somebody is trying to work out why the accounts do not reconcile with the job records, and the investigation costs many multiples of what the original correction would have — assuming it is found at all.

The asymmetry is what makes the function worth taking seriously. Every control that seems fussy at the point of entry is being paid for by something much larger further down, and every shortcut taken at the keyboard is a loan against a cost that lands on somebody else. A supplier invoice coded to the wrong job takes one keystroke to get wrong and can distort a job’s recorded profitability permanently, because by the time anybody questions the figure the invoice is a year old and nobody can remember which roof it was for.

Which is why the standard here is not speed and not even accuracy in the abstract. It is that the record matches the source, and that anything the source did not establish is visibly unestablished rather than quietly completed.

What data entry produces, and what holds each piece up A horizontal spine carries three labelled artifacts produced by data entry: fields that match the source, a flag on anything the source did not say, and a second look at what changes character. Beneath each artifact a vertical line drops to a second tier naming what holds it up — respectively keyed from the original, never from a summary, the gap raised at the keyboard instead of filled in, and the combinations that carry consequence known on sight. Across the foot of the diagram a separate band states the judgement this function does not make, which is whether the source material itself is correct. The diagram shows structure only and contains no figures. What this desk produces Fields that match thesource A flag on anything thesource did not say A second look at whatchanges character keyed from the original,never from a summary the gap raised at thekeyboard instead of filledin the combinations thatcarry consequence known onsight Every item above sits on the one below it. Outside this desk Whether the source material itself is correct
Each artifact on the spine has something underneath holding it up. That is the whole design: data entry is not an argument, it is a set of items that can each be traced to how they were arrived at, so that an error is caught where it is cheap rather than where it is visible. The band across the bottom is the part that stays outside the work.

Some fields change what a record is

Most of what gets keyed is ordinary business information. A small part of it is not, and the difference is defined rather than left to feel. One state’s definition, cited here as a worked example of the class rather than as a statement of national law, is in California Civil Code section 1798.81.5 — Reasonable security procedures, and what counts as personal information, which treats as personal information “An individual’s first name or first initial and the individual’s last name in combination with any one or more of the following data elements, when either the name or the data elements are not encrypted or redacted” — going on to enumerate identifiers of the kind that turn up on a subcontractor’s paperwork or a financing application.

Read the structure of that rather than the list. The trigger is a combination. A name on its own is not it. An account number on its own is not it. The two arriving together in the same record is, and that is a distinction a person can make at the moment of keying and essentially nobody can reconstruct afterwards from a database.

The practical consequence for this desk is narrow and concrete. It is worth knowing which fields, on which forms, produce that combination, so that those documents are handled deliberately — not held longer than the work needs, not copied into working files, not left in a mailbox as an attachment. What the desk does not do is decide what your company’s obligations are. Which rules reach your business, and what they require of it, sits with you and your advisers; the crm and job-management department page argues that side of it, and other states word all of this differently.

This describes how a definition affects the handling of documents at a data-entry desk. It is not legal advice, it makes no determination about which rules reach your company or what they require, and one state’s treatment is an example of a class rather than a national standard. Those questions belong with your own counsel.

Before this desk and after it

Data entry is one desk inside Administrative Support, and the limits that govern the whole department are set out there rather than repeated here. Getting the documents in and keeping them findable is a separate desk from keying what is on them, and cleaning up a set of records that is already wrong is a third thing again.

What owners ask about keying

What gets keyed, in practice?

Whatever arrives on paper or in a format your systems cannot read. Supplier invoices into the accounting file. Measurements and material lists off a takeoff. Homeowner details from a signed agreement. Crew hours from timesheets. Warranty registrations from a manufacturer’s form. The common feature is a document on one side and a set of fields on the other, and somebody having to move between them accurately while the phone is going.

How do you catch mistakes?

Mostly by removing the conditions that produce them rather than by checking afterwards. Keying from the original rather than from somebody’s transcription of it. Doing one document type at a time rather than switching. Never inferring a value the source does not give. Checking twice is the weakest control available, because the second look is done by the same person carrying the same assumption that produced the first error.

What happens when a document is unclear?

It gets flagged and left empty, not filled with the most likely value. This is the single most important habit in the function and the hardest to hold to, because a guess makes the record look finished and an empty field looks like unfinished work. An empty field is honest and somebody will resolve it. A confident wrong value is indistinguishable from a right one and will be believed by everybody downstream indefinitely.

Do you work in our systems or export things?

In yours. Data keyed into a parallel file and imported later is data that goes stale between the two events, and it introduces a reconciliation step whose only purpose is to correct problems the arrangement created. The whole point of the function is that a document arrives and its contents exist in the place your company actually looks.

Is any of this risky from a privacy point of view?

It deserves a straight answer rather than a reassuring one, and the section below sets out the specific distinction that matters. In short: some combinations of fields change what a record is, the difference is defined rather than a matter of judgement, and a desk that can recognise those combinations as they arrive handles them differently. The obligations themselves rest with your company, not with us.

Can you fix historical data as well?

That is a different job with a different shape, and it belongs to the migration and cleanup desk rather than this one. Cleaning up an existing set is about deciding rules and applying them in bulk; this function is about the next document, arriving now, being keyed correctly. Doing the second without the first means tidying a set that is still filling with errors.

Where the definition above comes from

Pick a job that lost money and trace one invoice

If it takes more than a minute to establish which roof a cost belonged to, the answer was decided at a keyboard nobody was watching.

Get in Touch Call 615-873-2843