ABSTRACT Red giant asteroseismology can provide valuable information for studying the Galaxy as demonstrated by space missions like CoRoT and Kepler. However, previous observations have been limited to small data sets and fields of view. The TESS mission provides far larger samples and, for the first time, the opportunity to perform asteroseimic inference from full-frame images full-sky, instead of narrow fields and pre-selected targets. Here, we seek to detect oscillations in TESS data of the red giants in the Kepler field using the 4-yr Kepler results as a benchmark. Because we use 1–2 sectors of observation, our results are representative of the typical scenario from TESS data. We detect clear oscillations in ∼3000 stars with another ∼1000 borderline (low S/N) cases. In comparison, best-case predictions suggest ∼4500 detectable oscillating giants. Of the clear detections, we measure Δν in 570 stars, meaning a ∼20 per cent Δν yield (14 per cent for one sector and 26 per cent for two sectors). These yields imply that typical (1–2 sector) TESS data will result in significant detection biases. Hence, to boost the number of stars, one might need to use only νmax as the seismic input for stellar property estimation. However, we find little bias in the seismic measurements and typical scatter is about 5–6 per cent in νmax and 2–3 per cent in Δν. These values, coupled with typical uncertainties in parallax, Teff, and [Fe/H] in a grid-based approach, would provide internal uncertainties of 3 per cent in inferred stellar radius, 6 per cent in mass, and 20 per cent in age for low-luminosity giant stars. Finally, we find red giant seismology is not significantly affected by seismic signal confusion from blending for stars with Tmag ≲ 12.5.