Liking relevance

A nice talk, and even though the subject (and the project) are known, I still found a few new things.

I think that you should relax when trying to speak. I know your english is good, so no need to emphasize your sentences.

I liked the move from simple elements to the more complex elements, but the same as Stefans talk, I cannot figure out what the target audience is. If it's people who have no idea what Solr is, or are just into this talk for the use-case, this talk will be very hard to follow.

For people with (some) solr knowledge, but have no idea about the project, you might try and elaborate your cases a bit more detailed (like the pivotting), and the reasoning why you went for the given solution.

But overall a nice presentation.

Thanks Jeroen for the talk.
It is very interesting to hear a core developer of a successful site like AutoTrack give some insight about how you pulled it off.
Maybe it would be nice to start with a little introduction about what Solr is, and which problem it solves.
And I don't really get your title, it's not a headline that sells the talk.
The practical examples were really nice. I think you need to find a way to increase the font size if the PHP code, this will not work in a bigger room. The font of the resulting queries is about the right size I think.
I would really be interested in a Workshop on setting up Solr and running the first queries, for example during the Dutch PHP Conference. You can take some shortcuts by supplying a simple schema and a data JSON file.

Having worked on parts of the Autotrack project myself, I cannot really judge how the presentation would be without this background info.

I think the project has some really interesting and challenging solution, so is a good candidate for a case study. A few tips:

- in some cases you refer to the autotrack website, in other cases to (simplified) raw query result data. I think the raw data can in some cases be helpful to show the limitations you ran into, but it might be a good idea to start each subject with a quick demo / screenshot of the functionality in the autotrack website. That helps people to connect a certain technical solution to the result for the end user, where the data alone might not always be enough for people new to the subject. That also fits with a case study, I think.

- you briefly explain there are multiple indexing strategies and that you use two, a full reindex nightly and incremental updates during the day. But you don't explain why.
This was eventually covered in an answer to a question at the end of the presentation, but it might be better to briefly touch this in your presentation since you mention two strategies.

- the tagging / excluding is a hard subject to explain, in my experience even to people with some basic Solr experience. I think it would help to spend a bit more time on explaining exactly what functionality you need, using the Autotrack website as a reference, before going into the technical details.
Also you could try to emphasize that you are basically 'picking and choosing' filter combinations for each facet, where the resultset itself always has all filters applied. I think once you get the concept clear, the technical explanation makes more sense. And even if the audience would still not be able to follow all the implementation details, they would at least have a good impression of what functionality you managed to create, which might be more important than the technical details (as you said, you could talk about that subject alone for hours to get it fully clear)

Really good talk on Solr. I've recently done a project with Solr, and now already I know a couple of things I can do better next time. Thanks for all the insight!

I really liked the talk and could easily follow even though I have never worked with Solr before.
A quick introduction to Solr could be handy in the beginning like Jeroen said.
But as soon as you started I could follow thanks to every example.

It has been said before but the font could indeed be a bit larger.
And when you switch from example to a view of the GUI it was really hard on the eyes ( black to white), at least on the front row.

Don't forget a 'Group By' example!

Overall a very good talk and it got me really interested in Solr!