Okay so keeping in mind your sample size is 28, benchmark is 14, average time spent with customers is 9.07, std dev is 1.86, std error is 0.35 which leads to a t-stat of -13.99. The hypothesis could be any of following:
P-value
5% critical value (one-sided upper tail test)
1.HO: m <14 (benchmark)
2.HA: m>14 (benchmark)
3. Test using p-value:
T-stat: remember coefficient-benchmark/ std error which is 9.07-14/0.35 =-13.99
P-value: 1.00
4.Decision rule at 5% significance level: reject null if p<0.05:
a)p-value >0.05 fail to reject
5.Conclusion 
There is insufficient evidence at the 5% significance level to conclude that the average time spent per customer is greater than 14 minutes.

P-value
5% critical value (one-sided lower tail test)
1.HO: m >14 (benchmark)
2.HA: m<14 (benchmark)
3. Test using p-value:
T-stat: remember coefficient-benchmark/ std error which is 9.07-14/0.35 =-13.99
P-value: 0.00
4.Decision rule at 5% significance level: reject null if p<0.05:
a)p-value <0.05 reject
5.Conclusion 
There is sufficient evidence at the 5% significance level to conclude that the average time spent per customer is less than 14 minutes.

P-value benchmark, any difference or not the same
5% critical value (Two-sided test)
1.HO: m =14 (benchmark)
2.HA: m≠ 14 benchmark)
3. Test using p-value:
T-stat: remember coefficient-benchmark/ std error which is 9.07-14/0.35 =-13.99
P-value: 0.00
4.Decision rule at 5% significance level: reject null if p<0.05:
a)p-value <0.05 reject
5.Conclusion 
There is sufficient evidence the average time spent per customer is not at benchmark at 5% significance level.

Confidence Interval benchmark (two-sided test)
1. HO: μ = b (14)
2.HA: μ ≠ b (14)
3. Test using 95% confidence interval:
(8.38, 9.76) lower bound and upper bound.
4. Decision rule: reject if 14 outside of the 95%  CI
a) 14 is outside the CI - reject 
5. Conclusion 
We are 95% confident that the average time spent per customer is not performing at the original benchmark of 14 being the time taken before the training was implemented. Instead, the average time post training is now between 8.38 and 9.76.

Confidence Interval benchmark (two-sided test)
1. HO: μ = b (14)
2.HA: μ ≠ b (14)
3. Test using 99% confidence interval:
(8.16, 9.98) lower bound and upper bound
4. Decision rule: reject if 14 outside of the 99% CI
a) 14 is outside the CI - reject 
5. Conclusion 
We are 99% confident that the average time spent per customer is not performing at the original benchmark of 14 being the time taken before the training was implemented. Instead, the average time post training is now between 8.16 and 9.98.

Hope this helps tanel :)