Every census result I have seen gives higher figures for JW menbership than their own publisher numbers: Canada, Mexico, Brazil, Australia, New Zealand, or any country that has a census which asks citizens to specify which religious denomination people belong to, sometimes by a significant margin. So in terms of self-identified affiliation, I wouldn’t be surprised if the worldwide figure would be over 10 million.
Let's unpack this a bit. I don't think the average Witness cares much about how the WT goes about deciding who counts and who doesn't count for purposes of their annual statistics. They're much more likely to pay attention to the rate of growth or decline.
The WT has been using the same metric for decades. I have very little doubt that the number of people who self-identify as Jehovah's Witnesses is larger than the number of publishers who have turned in a time slip at least once in the past six months.
In a sense, you can view this as a sign of honesty: If the WT were truly dishonest, they'd start counting everyone who attends the Memorial! But actually, an abrupt change like that would be very noticeable and would likely be more detrimental than not.
The 10 million figure you cite isn't particularly helpful if the goal is determining whether the WT is "cooking the books." If you could reliably compare the 10 million figure versus what that same figure was 1, 5, 10, and 20 years ago, you could start to plot trends on a graph. One would expect that trend to parallel official WT number of publishers. If there was a significant discrepancy, it would be strong evidence that the WT numbers aren't reliable.
In other words, the real doubt is the WT's honesty in revealing the true rate of growth or decline. The raw number of Witnesses isn't really an issue as there are many different credible ways of trying to arrive at the definition of what constitutes a Jehovah's Witness.