You were lucky but do not try this too often. Your 6BL7 was heavily used and this is why you were able to use it at around 70% of the datasheet specs. But this is not always the case. For short periods of time I guess it is fine (you stop the amp if you see that the current raises too much and that's it) but look at this 6SL7 graph. We would need around -0.7V for 3mA. Right?
Well, check one of my Fivre when measured. I have here 2 situations I want to show you. First, the tube is above 100% (according to datasheet) because it has 3mA at around -1V (30% above?) and it is also unbalanced internally. When I set on my amp -1V grid bias one of the triodes goes to 2.6ma while the other is around 3.7mA.
And you can see that for around -0.1V I have a variation of 0.2mA. This delta parameter can vary and I know it has a name, but I am a noob and forgot it. Doesn't matter, numbers speak for themselves. In short words, what I am trying to say is, never think "has to work with +2V of bias because the trimmer does not go so low or it worked once". And always check both channels, this Fivre on one channel might have 4mA and on the other 5.1ma but there are worse scenarios.
And another thing that I have learned during time while measuring these, is that it is no rule that a tube which measured 100% at a specific seller (if he tells the truth LOL) will be at 100% in our amps. Because they measure in specific conditions, taken from datasheets and measurement's device manual while we use them in totally different conditions. We would expect a linear variation with plate voltage? Well, I had surprises. Tube measured perfectly according to the datasheet was not as expected at 180V and so on. And this discussion might go more and more into what means 100%, what means matched pair and so on. For me nothing, except 2-3 persons I trust and from which I buy with confidence, the rest we measure and put in a box if does not work as expected or sell to people with automatic bias. I bought tubes measuring 70% from Langrex as "NOS", and electrically matched in tolerance of 15% and was told it's fine.
Later edit: For output tubes I use exactly the same method to determine the grid bias, I just look at "Triode connection graph" because outputs are triode strapped on the amp. So same stuff, just be careful to search for the proper graphs otherwise things might go crazy (this is valid only for Eternity).
Later edit 2: I went to people smarter than me
here and learned that the parameter I was mentioning is the gm (transconductance, this case ratio but anyway).