If you want to test whether an effect is significantly different from zero, you can construct a 95% confidence interval and check whether the interval includes zero. In the process, you get the added bonus of learning how precise your estimate is. If the confidence interval is too wide, you may need to collect more data.
