From tender to quote: why estimators lose days to GAEB — and how to get them back

Ask a window or door manufacturer where the bottleneck in their sales process is, and the answer is rarely "pricing." It is everything that has to happen before anyone can price: getting a messy tender into a form the estimating system can actually work with. That preparation routinely costs days per tender. Here is where those days go — and why almost none of it is real estimating work.

The hidden work before the first price

A typical tender lands as a GAEB file (often the older GAEB 90 or a DA XML variant), sometimes as a PDF, sometimes as an Excel export someone made from a GAEB file. Before an estimator can quote, they usually have to:

1. Open it — and hope the estimating software reads that particular format and phase.
2. Retype what won't import — positions, quantities, and long execution texts that got mangled or dropped.
3. Interpret each position — figure out which product, profile, or assembly each line item actually refers to.
4. Match it to the catalogue — map that interpretation to an internal article and a price.
5. Fill in the D84 — enter unit prices back into the exact structure the client expects.
6. Check it survives export — confirm nothing broke on the way back out.

Only steps 3 and 4 are estimating. Steps 1, 2, 5, and 6 are pure data handling — and they are where the days disappear.

Why it is so slow

Three things compound:

• Format fragmentation. Because several GAEB generations coexist, no single tool reads all of them cleanly. Every unsupported file becomes manual retyping.
• Volume. A single tender can run to hundreds or thousands of positions. Even a few seconds of human handling per line adds up to hours.
• Fragility. The return file (the D84) has to match the request (the D83) exactly. One misaligned position or encoding slip can make the whole bid unreadable to the client — so estimators re-check obsessively, because a silent failure is worse than a slow one.

The result: senior, expensive people spend the bulk of a tender doing work that requires none of their judgement, and quote turnaround stretches to the point where some tenders simply don't get answered in time.

What actually removes the time

The fix is not "estimate faster." It is to delete the data-handling steps so the estimator only does the part that needs a human:

• Read any format once. A parser that handles every GAEB generation and phase means step 1 and step 2 disappear — nothing gets retyped because nothing failed to import. This is what our open-source engine, pyGAEB, does.
• Propose the match, keep the judgement. Software can suggest which catalogue article and price each position maps to. The estimator reviews and corrects rather than building every line from scratch — steps 3 and 4 become review, not data entry.
• Guarantee a valid return. Generating the D84 straight from the original D83 structure removes the "will it survive export" anxiety of steps 5 and 6.

That is exactly the shape of the DatumOS platform: upload a tender in any format, and get back a priced, reviewable, send-ready quote in minutes — with a human in the loop for the pricing decisions that matter, and out of the loop for the retyping that doesn't.

The number that matters

When you strip out the data handling, the estimator's day stops being "one tender, if we're lucky" and becomes "several tenders, properly reviewed." That is the real return: not shaving minutes off pricing, but reclaiming the days that were never estimating in the first place.

Want to see it on one of your own tenders? Talk to us — you'll speak to the people who built the engine, not a sales desk.