I think that both Estibot and GoValue should be consulted, but as many others have posted you need to weigh the information and sometimes one or both will be totally wrong (sometimes low, often high).
It is important to realize what each stresses. For Estibot they place emphasis on how many TLDs of the domain are taken, popularity of the search term, the CPC and ad competition, etc. as well as previous sales. It depends on your potential purchaser if that emphasis is important. Estibot for these reasons really favours English (or other) dictionary words that are short and heavily searched. I have found they don't handle most "brandable" words as well (but that is clearly hard to do!), unless that word has already sold in one extension.
If using the new beta Estibot, I have found they have mistakes saying extensions are available when they clearly are not. As a result they make a domain value low.
For GoValue they really stress if that exact word in a different extension has sold for a high price. That is if xyzabc.com went for more than $25,000 then they will say that xyzabc.net .org and the new gTLDs are also valuable. They don't discriminate much between extensions which is a flaw. Estibot on the other hand seems to with country code and new gTLDs have an almost binary choice and is super harsh on some extensions in a way that does not seem warranted.
I would not discount a name because both value it low, IF you can look at the methodology and see why they valued it low, and feel that the name is still valuable.
I would not buy a domain ONLY because they both value it highly, but as others say, it warrants a closer look.
One value of the tools is the information they provide beyond the value. You can see comparable domain sales on GoValue that are not in the Namebio database in many cases. The search/ad statistics from Estibot are helpful.
In the world of AI, even current technology, automated tools could be much better than they are. Things like how well a domain matches the TLD, and how often two words work together, could be programmed much better. It would be good if a major AI strong company like IBM or Google built a new generation AI tool. I think the GoValue competition will force Estibot to evolve and improve, and that is good. Right now I view the two as roughly equally good/bad.
Keep in mind that values are not static. As has been mentioned, Estbot at least sometimes adjusts new values to sales prices, which is justifiable but means if you buy a domain at a great price Estibot may soon say that bargain price is the value of the domain (if you don't manage to keep it private). GoValue have not yet at least adjusted worth of some of the big sales last month - e.g. they still say home.loans is worth $1600 although it sold for $500,000. I can't prove it but I suspect that GoValue prices are influenced by BIN prices on the Afternic network.
For what its worth (not much perhaps!), I have valued my entire portfolio using both Estibot and GoValue. There are 184 domain names in the portfolio, and if I count <$100 as $0 for GoValue estimates (perhaps harsh) the total GoValue worth of the portfolio is $94,311 and the total Estibot valuation is $84,500. My BIN prices for the same portfolio total $22,238, and I would probably on average happily accept about half of that or a bit less. Note this is updated from a previous post a few weeks ago, and this one includes only those domains currently in my portfolio, not ones I have sold or looked at - in total I have done both Estibot and GoValue on about 400 domain names and may do an analysis on them all at some point.
While it is tempting to seek a factor like 5% or 10% of Estibot value, I don't think that it is possible to give a useful single factor. You should not count on always getting 5% of Estibot, but nor will it always make sense to sell at less than full Estibot/GoValue. Some names Estibot says are $0 I would not sell for less than $100, many that both say are the order of $1000 or more I would happily sell for $50.
I think both tools do best on domain names in the $$$ range. Estibot should not try to distinguish $10 from $80 and simply say <$100 like GoValue. GoValue should not give a precise value like $1231.
GoValue (in beta) seems to have now settled down to more steady prices - it jumped around more in the first weeks after its release. Overall I would rate GoValue at B+ and Estibot at B, both worth using with caution, but both could be improved. Different weaknesses, but overall about equal in overall value. The fact that GoDaddy sell domain names is a conflict of interest, as others have pointed out, although I suppose same as art auction prices predicting a range for a piece of art going to sale.
Hope some of this is helpful.
Bob