Get answers fast

Each one of those tabs has a list at the top showing what's already configured and a form below for adding new entries. Click Add and fill in the fields - they all default to common starting values, so you can save fast and refine later.

Edits go through the same form: click "Edit" on any row to populate the form, then save. The cost engine picks up changes immediately on the next quote.

Until your domain is verified, customer-facing emails go from our shared sandbox sender (your customers see "via Quotram" in some clients). After verification, every email comes from your domain with your shop name as the display name.

We also surface a DMARC check on the same page once your domain is verified. DMARC is optional but strongly recommended - Gmail and Yahoo's bulk-sender rules treat unauthenticated mail as suspicious without it. The settings page shows a recommended starter record (v=DMARC1; p=none;) you can paste if you don't have one yet.

The logo renders on the customer-facing PDF, the portal page, and transactional emails. Transparent backgrounds (PNG / SVG) generally look better than JPEG against the letterhead.

If you upload something larger than 2 MiB, the picker tells you the size before the upload starts so you can compress and re-pick.

When we extract data from a drawing - material, dimensions, hole counts, finish - we show a small confidence pill next to each field. The model is instructed to score itself conservatively: if it's looking at a smudged title block or an ambiguous note, it lowers the score rather than guess.

On the part-review page, low-confidence fields highlight in yellow (0.5–0.7) or red (below 0.5). Click any field to edit and override. Your correction is saved and used to improve the prompt over time.

Misses happen on drawings that don't follow conventions (no title block, hand-written callouts, scanned poorly). Open the part, click into the missing field, type the value, and save.

Behind the scenes we record your correction so future extractions on similar drawings improve. You don't need to re-upload - the part keeps its original file and just gets the corrected fields.

The quote review page lists every cost line by category: material, cut, form, weld, finishing, plus markup lines for overhead and margin. Each line shows the calculated cost; for time-driven lines it also shows the cycle time, setup, and rate that produced the number.

Hovering a line surfaces the exact formula the cost engine used - for example, "(perimeter 42.0 in / 80 in/min + 8 pierces × 0.5 s) × $120/hr × 1.0 cal". Every dollar in the quote can be traced back to your shop config.

You can override any line: click the pencil, enter your own number, optionally add a note explaining why. The override survives margin and quantity changes; the calculated number is preserved alongside so you can revert later.

When a customer clicks "Request changes" on the portal, you get an email with their note, and the quote's status flips to revision_requested. Open the quote in the app, make whatever changes the note asks for, and click Send. The customer receives a fresh email with the revised quote attached.

The portal link they have stays the same - it just renders the latest revision when they open it.

Hard bounces come back from the recipient's mail server when an address doesn't exist or is rejecting mail. We capture the bounce on the customer record and the next attempt to send to that customer fails fast with a clear error - re-sending to a known-bad address damages your sending reputation.

Fix: update the customer's email under Customers → [their record] → Edit, then open the quote and click Send again. The new address gets a fresh delivery attempt. Soft bounces (full inbox, temporary rejection) don't lock the address - those auto-retry.

The cap catches two failure modes: a misbehaving retry loop blowing through hundreds of sends, and accidental bulk-marketing patterns that would torch the shop's sender reputation with mailbox providers. When you hit it, the quote stays in draft and the limit resets at 00:00 UTC.

If you're a high-volume shop that legitimately exceeds 100 sends/day, email support@quotestream.com - we'll raise your cap.

Every shop's real-world numbers drift from the rule-based starting point - operator habits add minutes, supplier deals shift material costs, etc. After 10 won outcomes for a given operation type, Quotram computes a correction multiplier from the spread between your estimate and your actual paid price, then applies it to future estimates.

The multipliers are visible at Settings → Calibration. You can pause a multiplier (lock it at 1.0), reset it to 1.0, or trigger an immediate recompute with the "Run now" button. Below 10 outcomes the multiplier stays at 1.0 - too few data points to be trustworthy.

Calibration applies before markup, so a 1.15 multiplier on material cost means the material line goes up 15% and overhead/margin compute off the higher number - which matches owner intent ("I want 35% margin on whatever it actually cost me").

Card 1 - Calibration vs. baseline. Shows whether the calibration adjustment moved this estimate up or down vs. the rule-based baseline, and the sample size behind the multiplier. When you don't have 10 outcomes yet for an operation, it shows "data accumulating, currently at N/10" instead.

Card 2 - Win rate by margin tier. Sparkline of your actual win rate across margin bands (under 30%, 30-35%, 35-40%, 40-45%, over 45%). Helps the estimator see "we win most jobs at 35%, drop off above 40%" so they can place this quote consciously.

Card 3 - Customer pattern. Auto-detected pattern from this customer's quote history (negotiator, fast-mover, ghoster, steady, etc.) plus typical acceptance time. Hidden when the customer has fewer than 3 quotes - patterns from 1-2 data points are noise.

/analytics has six widgets covering the shop's overall pricing and win/loss reality - all rolled up nightly from a materialized view of quote outcomes. Use it to ask shop-level questions: "Are we leaving margin on the table? Where does our win rate fall off?"

/customers/[id] is the per-customer drill-down. Health pill (active / cooling / lost), communication-pattern pills, suggested next action, and the full quote history. Open it before a sales call to see what you know about the customer at a glance.

Cancellation is at-period-end by default - you keep access until the end of your current billing cycle, then the account locks. Re-subscribing later is a fresh Checkout session via the same Manage button.

If you cancel during the trial, no card is charged and access ends at trial expiry.