It is not surprising to achieve similar things by logistic and SVM. No need to worry. This is true in many cases. Actually, logistic can be even better sometimes. SVM is fancier method for sure, which makes logistic regression seem too plain. However, there is no model that is bond to win in any cases. The best performance is given by the method that is the most suitable for the data. SVM uses hinge loss function with <bblatex>l_2</bblatex> regularization, which is not that much different from logistic regression, with <bblatex>l_2</bblatex> penalty. And in some circumstances, you may find logistic is more robust to noisy variables.
The reason why you didn't achieve that high precision might be the features you used and the parameters in the model. These steps are very important for the performance, and that's where a data scientist is valuable. Such adjustments are arts rather than plain mathematical knowledge. So don't be disappointed. Google pays more than $300,000 for its top data scientists. That is not a free lunch :-)
