The architect is wrong with the estimates. And now?

very often through unplanned encouragementn...

that hits the mark anyway :D

It’s not completely clear to me. But you probably can’t make it any cheaper at halftime. I also assume that the shell is already completed, right? If you haven’t planned any or enough buffer, I don’t have any good advice either. Except for additional financing. Or postponing the outdoor facilities to a later time.

Shell construction is the only trade that can really be calculated easily. And also secured with offers. Everything else automatically incurs additional costs, very often due to unplanned upscaling...

But that's exactly the question, is the comparison basis cost estimate, cost determination, or offer?

You can secure any trade with an offer, not just the shell construction.
If you leave out upscalings (or did you really mean upscalings ;) ), you can determine quantities and thus costs for almost all trades.
I can only report here from my own experience and that of my known environment. Often the shell construction became more expensive. The first reason was the increased prices which in our region for shell construction felt like they rose monthly. The second was often the earthworks, which were mostly included in the shell construction bill of quantities. Because those costs also constantly increased. The third was then the statics or the amount of required iron and then also the steel price. A large share here is also the basement, I don't know whether the OP is building with or without it.

And yes, anyone building with an architect should still allow for a buffer during the cost estimation phase.

When building with an architect, you don't do it without a tender before awarding the contract. Large deviations between estimates—unless you are rebuilding the roof of Munich’s Olympic Stadium—indicate that the architect is either a beginner or practically already retired, meaning they have little or outdated experience. A one-third cost increase (in a single-family home, not at BER or the Elbphilharmonie) doesn’t just happen by magic or because the collective bargaining for masons was suddenly taken over by the pilots' union. Nor is it due to fluctuating daily prices on the spot market for construction steel. An architect does not get paid just to draw a floor plan and send it to the building authority with the structural engineer’s comment; budget adherence is part of the service, not a cocktail cherry on top.

Often, when building with an architect, you are also building a house that has never been built exactly like that before. Whereas prefab or general contractor houses have often been built many times before with only minor deviations. There, it is better known what is needed in detail.

These are all myths in bags (and beware: starting 12/1 myths may only be sold in pouches!). General contractors nowadays make their main business with "individual planning." And the difference between a WC door hinged on the left for Meier and on the right for Müller doesn’t make that much of a difference either.

And think carefully about whether you want to recoup that with the kitchen and tiles,

That would be stupid not only because the quality of the design parts significantly contributes to the perceived value of the house. But the butter runs tend to be more about windows than tiles ;-)

You certainly don’t build with an architect without tendering before awarding the contract. Large deviations between estimates – as long as you’re not rebuilding the roof of the Munich Olympic Stadium – indicate that the architect is either a beginner or actually already retired, meaning he has little or outdated experience...

I repeat my question. Cost estimate? Cost determination? Or tender phase for comparison.
The OP is talking about 30% for the shell construction, not the total costs.

Estimates are not made based on individual quantities. Instead, the architect takes the gross floor area x estimated price from experience-based values.
I do not want to rule out that there are occasional estimates that are +/- 5%. But one should take the trouble to browse this forum or the rest of the internet a bit more and do some research. It is more the rule than the exception that there is a 20-30% difference between the estimate (!) and the award. In shell construction. Furthermore, which may now be speculation, the supposed shell construction offer may sneak in one or the other work that actually belongs elsewhere? Keyword ancillary construction costs, wasn’t the garage/carport calculated separately, groundwork/earth removal in shell construction or actually calculated in ancillary costs. All things that are not immediately obvious to a layperson given the not few line items in a bill of quantities.
Admittedly, 30% is already at the upper limit.

Estimation is not made based on individual quantities. Instead, the architect estimates by multiplying the enclosed living space by an estimated price derived from experience.

Correct.

I don't want to rule out that there are sometimes estimates that are +/- 5%.

That also depends on differentiating classes and not transferring experience values carelessly between a bungalow and a mansion (apples and oranges, basically).

But one might take a little more trouble to browse through this forum or the rest of the internet and do some research.

Better not – as often as tile sizes and heat pumps are changed mid-gallop here, interior door decisions remain open as long as possible, and partition walls are replanked, that would only provide an excellent basis for finest milkmaid’s calculations :-)