Quantitative interpretation of proxy temperature records, especially during the time interval of small climatic changes such as the Holocene, can be compromised by inherent uncertainties in the proxies. Constraining and understanding the sources of the uncertainties may help to reconcile the mismatch, in terms of amplitude in temperature trends and variability, between model outputs and proxy records. Here, we report on two complementary pieces of our systematic efforts in quantifying proxy uncertainties. We analyzed multiple sea surface temperature (SST) proxies on three short sediment cores from the same multicoring deployment. Our results allow to quantify and attribute the errors to the instrument, the work-up procedure and the spatial heterogeneity of proxies in sediments, providing some insights into how to best control for the noise sources. In the case of TEX86L and TEX86H, the spatial variability is as large as the downcore variability, suggesting that any interpretation of temporal changes in these proxies is highly uncertain at this study site. Geochemical proxies often suggest different amplitudes of glacial-interglacial (G-IG) SST variations for a given site. In many cases, these differences in G-IG amplitude are explained by invoking differences in seasonal production and habitat depth of the source organism of proxies. To further look into this issue, we compile published SST records from global sites with multi-proxy records. Analyzing the modern and simulated glacial seasonal and vertical temperature distributions allows us to test simple explanations such as systematic season and depth habitat difference. Comparing the modern-day spatial relationship between proxies (based on published global surface sediments data) and downcore / temporal relationship between proxies suggest that the modern spatial relationship might not always hold over time.