The cost of a bad RFI
An unanswered RFI doesn't cost you the day you send it. It costs the day the crew arrives to build the thing you asked about, finds the conflict still open, and goes somewhere else — plus the day you spend resequencing, plus the day the sub bills you for. None of it gets logged as a delay, which is exactly why it's the margin leak nobody bills for.
The fix isn't writing more RFIs — it's writing them so the architect can answer in one pass.
The five fields a great RFI always has
1. Drawing reference + revision
Bad:“The lintel callout doesn't match.” Good:“A-301, rev 3, grid 4 — the lintel callout L-2 conflicts with the shop drawing HM-08-410.” Architects need exact coordinates to look up the right detail. If the rev is wrong, you'll get a non-answer and bounce back two weeks later.
2. The actual question, in one sentence
Bad:“Please clarify lintel sizing on grid 4.” Good:“Should grid 4 lintel be L4×4×¼ (per A-301) or L6×4×⅜ (per HM-08-410)?” Specific options force a binary answer. A vague question gets a vague response.
3. Schedule + cost impact, your honest estimate
Adding “Schedule impact: 2 days · Cost impact: $1,200 (mat'l) + 6 hrs labor” signals to the architect that an answer is on the critical path. They'll prioritize it. Estimating low and getting hit later is worse than estimating high; if you're wrong, the CO covers it.
4. A photo or marked-up drawing region
Crop a region from the PDF, circle the conflict, attach. The architect's answer is faster and you've created a paper trail tied to a specific visual artifact. If a dispute comes later, the photo is gold.
5. Proposed answer (when you have one)
Don't hide your read. If you think L6×4×⅜ is correct, say so: “We propose L6×4×⅜ to match shop drawing.” Architects appreciate a recommended path; many will simply confirm. The ones who push back will give you better reasoning than they would have unprompted.
The four mistakes that drag projects sideways
1. Burying RFIs in email.If your RFI is in line 4 of a long thread, it'll get lost. Every RFI deserves its own log entry with a number, status, and timestamp. The log is the source of truth at closeout.
2. Asking questions you could answer with a spec search. Spec sections are searchable; check there first. Architects respect contractors who do their own homework.
3. Letting RFIs sit answered but unimplemented. An “answered” RFI hasn't actually closed until the field installs the answer and the CO (if any) is in. Track three states: open, answered, closed.
4. Not chasing.There is no code or contract default to fall back on — A201 has the Architect respond “with reasonable promptness” and “within any time limits agreed upon,” which means the clock is whatever your contract or the architect's own RFI procedure sets. Find that number, print it on every RFI you send, and chase in writing before it expires. An RFI with a response time nobody agreed to is one nobody owes you.
What a good RFI log looks like
For each RFI: a sequential number, subject, drawing reference, the question, your proposed answer, schedule impact (days), cost impact ($), date sent, date answered, current status, attachments, and the email thread it came from (if any).
On a busy project you'll have dozens. The log is what keeps the field honest about what's built, the office honest about what's billable, and the owner honest about what was approved.
How neuroBLDR handles RFIs
Each project gets a dedicated RFI log connected to the project brain. The RFI Agent drafts subject + question from a rough note (or a voice memo). Start from a drawing region when you want that source context attached. Photos and marked-up regions can be clipped from the PDF viewer. RFIs link to change orders so cost impact doesn't fall through the seam.
The Risk Digest (Pro tier) runs every Monday morning and flags the RFIs approaching or past their SLA — 14 days by default, using the same aging the RFI list shows in-app, so the digest and the screen never disagree. By Tuesday you've sent the chase emails and nothing has rotted in the inbox.