Unexpected forecast results using forecast() or ARIMA / TBATS models

Question

I try to forecast sales using various methods, but each time the result is not satisfactory. When using forecast () function or ARIMA / TBATS models, I always get the same values ​​(12 periods in the future).

Below is an example of 2 products with historical sales from 2 years (per month). In 1 case, the result of the forecast is unrealistic (any method - constantly growing values despite "fluctuations" in historical data). However, in 2nd product, the forecast result is the same value for each period.

I was trying simple forecast function and ARIMA / TBTS models. Always the same result. What is the case? What am I missing? It's impossible to do forecast from this kind of input data (no repeatability). Should I try with more observations?

Product 1 (constantly growing value)

Product 2 (constant value)

Product 1 input data - historical sales

Product 2 input data - historical sales

    ns <- ncol(b_TS) h <- 24
b_sales_fcast <- matrix(NA,nrow=h,ncol=ns)
for(i in 1:ns)
  b_sales_fcast[,i] <- forecast(b_TS[,i],h=h)$mean

Thanks in advance for any reply.