I had occasion to revisit Data Guy’s January 2018 Author Earnings Report this week. And I found Data Guy has quietly slipped in new numbers for 2015 and 2016 US ebook market.
Said Data Guy on May 31 this year, slipped into the comments thread, four months after the main report went live,

US ebook sales totaled roughly 545 million units for the last 12 months, up from the 510 million & 525 million in 2015 & 2016.

Whoa, hold on, Data Guy. Will that be the same 2016 for that, in the February 2017 Author Earnings Report, it was clearly stated US ebook unit sales amounted to 487,298,000?

Data Guy has repeatedly told us not to compare the old Author Earnings figures from 2017 (and prior) with the new numbers from January 2018, to avoid awkward questions about why the latest BookStat / Author Earnings numbers (“not quite 100%” accuracy but “pretty close,” adding “90%”) bear no relation to the old Author Earnings numbers, which at the time were claimed to be accurate to “within 6%… and most of the time to within 2%.”
Now it seems that not only are we not to compare the old and new numbers, but Data Guy is rewriting the old numbers on the fly, and we are expected to take these new numbers at face value.
The latest, out of the blue assertion (May 31, 2018) that US ebook sales in 2016 totalled 525 million is based on what evidence?
The latest, out of the blue assertion that the 2015 US ebook market unit sales totalled 510 million is based on what evidence?
Data Guy stated very clearly in the January 2018 Author Earnings Report that BookStat only began counting sales based on its new, supposedly super-accurate methodology, in April 2017.
So where do these revised numbers for 2015 and 2016 come from?
How can we possibly take seriously the BookStat / Author Earnings numbers when Data Guy can’t make up his mind from one day to the next what the real numbers are and plucks historical numbers out of thin air to make his latest numbers more plausible?
Earlier this year I raised the issue of the missing hundreds of millions of digital library downloads Data Guy chooses to ignore in his assessment of the US ebook market.
Just this month OverDrive confirmed 2018 is on target to be another record-breaking year, with combined ebook and audiobook downloads expected to hit a quarter billion by end of year. None will be recorded by Data Guy.
Earlier this year the Audio Publishers Association valued the US audiobook market at $2.5 billion. Data Guy’s valuation?
Data Guy in the January 2018 Author Earnings Report, derived from the BookStat figures, said,

During the last three quarters of 2017, we recorded … $490 million in individually tracked audiobook sales … While this is not quite 100% of online sales during the period, it comes pretty close.

Extrapolating Data Guy’s nine month tally for the full year would give us $653 million, a far cry from the $2.5 billion cited by the APA.

Audio Publishers Association says 2017 US audiobook market worth $2.5 billion – more than double Data Guy’s BookStat valuation. Who are we to believe?

It’s not remotely plausible that difference was made up by off-line audio sales of CDs which Data Guy cannot track.
It’s not even plausible that most of that difference came from the digital libraries that, likewise, Data Guy cannot track.
By coincidence I addressed the many issues surrounding Data Guy’s numbers in an extensive report published here on TNPS on 29 May 2018.

GDPR. The Facebook scandal. Amid mounting privacy concerns, Data Guy continues to sell author’s private sales data to Big Pub. But is it even accurate?

Data Guy was aware of the report (Mike Shatzkin raised it with Data Guy directly) but Data Guy declined to offer any public response.
Now we find that, just two days later on 31 May 2018, Data Guy was busily rewriting history, tucked away in the comments thread of a four month old post, telling us that in 2016 the US unit ebook market was not 487,298,000 as we had been clearly told in February 2017, but had mysteriously climbed to 525 million.
Oh, and by the way Data Guy has also changed the dollar value for the 2016 US ebook market.
On May 31 2018 Data Guy told us that the dollar value for ebook sales in the US was down about 7%, to just over $3.04 billion. But that would put the 2016 value at $3.27 billion, significantly higher than the $3.177 billion we were told in February 2017.
While we value the concept behind Author Earnings in offering us broad insights into the US ebook market that would otherwise remain cloaked in mystery, and welcome the fact that this data is offered free of charge,
a) the continuing problems matching up the numbers from one report to the next,
b) the repeated exhortation not to compare the data in the old and new reports, and
c) the clear and calculated rewriting of the historical numbers when convenient,
begs the question what value the Author Earnings reports really have for us.
One can only hope the paying clients of BookStat are getting a more rational selection and a more considered and consistent set of numbers than we are.