NEWVectors or files. Pick a path.Start →
    Back to DiagramsRetrieval Mechanics

    A Nine-Second Day Disappears Into a Two-Second Average: What Temporal Windows Show That a Mean Hides

    Mixpeek writes every retriever execution into mxp_retriever_executions in your own namespace, so a retriever can read your own telemetry. Grouped by day, one window on 2026-09-04 averaged 9,397 ms against neighbours near 2,500, on only 10 executions. Averaged across the full period that day contributes almost nothing and disappears; bucketed by day it is the tallest bar on the chart.

    A portrait diagram showing the temporal retriever stage grouping 280 real retriever executions into six day windows. Five bars sit near 2,000 to 4,500 milliseconds average duration; one bar for 2026-09-04 is rendered in rose at 9,397 milliseconds average, far taller than its neighbours despite carrying only 10 of the 280 executions. A dashed divider marks 2026-09-05 and 2026-09-06, both absent because nothing ran those days. A code panel shows the attribute_filter into temporal pipeline and notes that the windows live in stage_statistics metadata, not in documents. A footer reads 280 documents in, 0 skipped, 6 windows, 7.22 ms of stage time.
    Mixpeek writes every retriever execution into mxp_retriever_executions in your own namespace, so a retriever can read your own telemetry. Grouped by day, one window on 2026-09-04 averaged 9,397 ms against neighbours near 2,500, on only 10 executions. Averaged across the full period that day contributes almost nothing and disappears; bucketed by day it is the tallest bar on the chart.

    Mixpeek writes every retriever execution into mxp_retriever_executions in your own namespace. It is a collection like any other, which means a retriever can read it. This diagram is the temporal stage grouping 280 of those executions into day windows and averaging their latency.

    September 4th is the window worth looking at: 9,397 ms average against neighbours near 2,500, and only ten executions behind it. Averaged across the whole period that day contributes almost nothing and disappears into the mean. Bucketed by day it is the tallest bar on the chart, and it is exactly the kind of thing a single rolled-up average is built to hide.

    The 5th and 6th have no bar because nothing ran. temporal emits the windows the data actually occupies and does not zero-fill, so a missing bar is a real gap in activity rather than a quiet day rendered as zero. That distinction matters before plotting the output anywhere that assumes a continuous series.

    One thing about this stage catches people building with it for the first time: the windows live in stage_statistics.stages.<name>.metadata.windows, not in documents. include_documents controls whether the source rows pass through to documents; with it false, documents is empty and output_count reads 0, which looks exactly like a stage that did nothing. It had already done all of it. A reduce stage's product may not be documents at all, so its metadata block is worth reading before concluding it produced nothing.

    Getting to this collection at all needed a second stage first. mxp_retriever_executions carries vector_indexes: [] and no vectors, so feature_search cannot source from it. attribute_filter can, since as a first stage it fetches directly rather than requiring a prior vector search. Running it that way, roughly a third of otherwise identical runs came back with zero documents; every one of those was discarded as unmeasured rather than counted as a result, and the numbers on this diagram come only from runs that actually delivered documents.

    Every number on the image traces to one execution against col_60783f4a99 in ns_cbb394f1a5, skip_cache true: 280 documents in, 0 skipped, 6 windows, 7.22 ms of stage time.

    Where this diagram appears

    Run this on your own data

    Mixpeek turns video, images, audio, and documents in your object storage into searchable, timestamped results through one API.

    Search your own data

    Related diagrams

    A portrait diagram in two rails. The headline reads Filters exclude, what ranking would have shown, with the subline: a pre_filter drops rows before ranking is ever shown to you, useful, and it hides what it removed. A search bar holds the query yellow armchair, annotated: three of the top five are out of stock. Rail A, badged no filter, shows five armchair photos in rank order with prices and scores: 872 dollars at 0.080, 515 dollars at 0.065, 180 dollars at 0.060, 451 dollars at 0.060 and 506 dollars at 0.057; the second, third and fourth are greyed and stamped dropped, and the count chip reads 5 of 5. Rail B, badged pre_filter with in_stock equals true, shows the two survivors at 872 and 506 dollars followed by three new armchairs at 123 dollars 0.048, 446 dollars 0.047 and 606 dollars 0.046; its chip reads 2 kept plus 3 new. A bold line states that three of the top five were out of stock and all three were removed from the response entirely. A dark code window shows the execute request with the filters line in_stock true highlighted, and skip_cache true. The footer prints mxp.co/d/filters-exclude and the words real query, real ranking, mixpeek.

    Filters Exclude What Ranking Would Have Shown

    A portrait diagram in two rails. The headline reads Scope decides, what can be found at all, with the subline: the same question, asked of the whole catalog and then of one category. The search bar holds the query something to keep me warm outside, annotated: the query names no category. Rail A, badged whole catalog with the note semantic reach across every category, shows five photos in rank order: a lit tent at night for 543 dollars at 0.062, a snowy forest scene for 564 dollars at 0.060, a steaming pot on a stove for 444 dollars at 0.059, boots on a textured rug for 203 dollars at 0.052, and a tent under a starry sky for 100 dollars at 0.048; the second, third and fourth are greyed and stamped dropped, and the chip reads 5 of 5. Rail B, badged scoped with category equals tent and the note only one category can answer, shows the two surviving tents followed by three more tents at 592 dollars 0.046, 73 dollars 0.039 and 554 dollars 0.029; its chip reads 2 kept plus 3 new. A bold line states that unscoped, the index answered with tents, backpacks, rugs and pans, and that scope is a decision about what is allowed to be right. A code window shows the execute request with the filters line category tent highlighted. The footer prints mxp.co/d/scope-changes-candidates.

    Scope Decides What Can Be Found at All

    A portrait diagram in two rails. The headline reads A price ceiling, is not a price sort, with the subline: a price ceiling removes the expensive matches, it does not promote the cheap ones by price. The search bar holds the query premium floor lamp, annotated: premium is a word, not a number. Rail A, badged no ceiling with the note best matches, any price, shows five photos in rank order: floor lamps at 729 dollars 0.097, 299 dollars 0.093, 908 dollars 0.092 and 322 dollars 0.089, then an armchair at 451 dollars 0.086; four are greyed and stamped dropped, only the 299 dollar lamp survives, and the chip reads 5 of 5. Rail B, badged pre_filter with price less than 300 and the note best matches under the ceiling, shows the 299 dollar lamp first at 0.093, then lamps at 230 dollars 0.071, 104 dollars 0.068, 129 dollars 0.063 and 38 dollars 0.059; its chip reads 1 kept plus 4 new. A bold line states that rank order inside the filter is still semantic, so the cheapest item lands fourth. A code window shows the execute request with the filters line price lt 300 highlighted. The footer prints mxp.co/d/numeric-filter-reranks.

    A Price Ceiling Cuts the Set and Leaves the Order Semantic