I have been critical of way many newspapers display reported coronavirus deaths per day, but this graph from NBC News has important advantages over others.
For one thing, the NBC graph does a nice job of pointing out that what appear to be "daily" numbers in the news can, in fact, be a collection of death records that happened to be dumped into a single day. The very conspicuous spike in "national" deaths on April 14 was because on that day New York City officials decided to add 3778 "probable" deaths from some unknown previous days or weeks. Why? Because April 14 was the first day the CDC guidelines said probability was close enough for government work (rather than testing or a death certificate). Ever since then, the death statistics are no longer comparable to previous dates or between states (because some states include probable deaths and others are more cautious).
Another big spike on "national" deaths happened on June 26, when The New Jersey Department of Health added 1,854 "probable" deaths that were thought likely to be have been related to COVID-19 because "COVID-19 was listed on death certificates as a possibility or because the person had symptoms (very similar to pneumonia and/or flu) but was never tested.
Because New Jersey opted to add 1,854 probable old deaths to the count on June 26, the widely-cited "national" 7-day moving average of daily deaths shot up by more than 200. It remained very high until enough days had passed to drop the June 26 fluke out of the average.
The average number of deaths over the previous 7 days fell from 1838 on May 1 (when reopening began) to 952 on June 1 and then to 719 on July 1. By July 11 it was down to 617. But three days later it was up to 757. Many reporters use such 7-day averages to paint an ominous escalation in national deaths over the past few days. But the bad news has been highly localized and very short-term.
Another advantage of this simple NBC graph, then, is that it does not add a line showing a rolling 7-day average. Too many statistical innocents have gotten into the bad habit of speculating that a brief rise in the 7-day average (like the one we had after June 26) constitutes an ongoing trend that can be assumed to continue climbing higher and higher. To call such a conclusion "premature "would be excessively kind.