Right--choosing the right tool is an art form. See, for example, the difference between these two.

MDLP (Magnus Ducatus Lituaniae Project - BGA analysis project for the territories of former Grand Duchy of Lithuania)

Least-squares method.

Using 1 population approximation:
1 German_V @ 3.658
2 Austrian @ 5.041
3 CEU_V @ 5.778
4 Hungarian @ 6.210
5 Welsh @ 7.777
6 CEU @ 7.793
7 Croatian @ 7.840
8 Bosnian @ 7.973
9 Serbian @ 8.016
10 Slovenian @ 8.326

Dodecad Africa:

Pct. Calc. Option 2

1 Tuscan 76.60%
2 French_Basque 8.93%
3 North_Italian 4.09%
4 Druze 4.05%
5 Jordanians 3.07%
6 TUNISIA 2.15%
7 Biaka_Pygmies 1.09%
8 North_African 0.01%
9 Bantu_S.W._Ovambo 0.00%
10 Morocco_Jews 0.00%


And EUTEST:

Pct. Calc. Option 2

0 Unable to determine 0.02%
1 English 49.42%
2 NL 11.84%
3 Orcadian 9.86%
4 IE 7.94%
5 Lezgin 7.40%
6 Scottish 6.74%
7 Udmurt 4.91%
8 West_&_Central_German 1.85%
9 GE 0.01%
10 Erzya 0.00%

It's clear that the choice of tools affects the outcome, and the one should experiment and find the tool that best fits the individual's ethnic background. In my case, a tool that uses a broad sample of European populations is superior to one that focuses on Lithuania (MDLD) or on Africa (Dodecad Africa).

It's worth bearing in mind this disclaimer from the developer of the Harappa World Project:

Do note that the admixture components do not necessarily represent real ancestral populations. Also, the names I have chosen for the components should be thought of as mnemonics to ease discussion. I chose them based on which populations in my data these components peaked in. They do not tell anything directly about ancestral populations. The best way to look at these admixture results is by comparing individuals and populations.

Jim