Energy storage systems (ESSs) can help to reduce the intermittency and uncertainty of renewable energy supplies in power systems. ESSs are critical components of renewable-rich standalone microgrids (SMGs) to balance power generation and load demand, which is referred to as reliability. To achieve the same level of reliability as conventional power systems for renewable-based SMGs, significant investment in ESSs is required. However, due to the high investment costs of ESSs, the installation of large ESSs will not result in an affordable solution for achieving renewable SMG at the required reliability. As a result, this paper proposes a new sharing concept for ESS, namely energy storage as a service (ESaaS), to be implemented across microgrids as a low-cost alternative for improving reliability. In the proposed ESaaS concept, microgrids can use ESS from an ESS provider as required for different timeframes such as monthly, weekly, or daily, depending on the renewable resources and load profile characteristics. In this paper, the use of ESaaS is investigated over a range of timeframes for a 100 % renewable-based SMG with photovoltaic (PV), wind turbine (WT), and ESS. The SMG reliability is evaluated using Monte Carlo simulation both before and after the ESaaS strategy has been implemented. To evaluate the ESaaS affordability in improving the reliability of an SMG, this paper proposes the criteria of marginal cost of reliability, which indicates the rate of additional investment amount per percentage of reliability improvement. The marginal cost of reliability combines the economic and technical aspects of ESaaS in one simple criterion for effective decision-making among investment strategies such as different timeframes of ESaaS or permanent ESS. The simulation results show the ESaaS based on daily contract results in a lower marginal cost of reliability for the case study. To validate the effectiveness of the proposed ESaaS approach using marginal cost of reliability, the levelized cost of electricity (LCOE) is also calculated for different strategies of reliability improvement. The results confirm that the lowest LCOE is obtained using the strategy that provides the lowest marginal cost of reliability for the case study. In addition, a sensitivity analysis is performed to assess the difference in marginal cost of reliability under various uncertainties associated with the installed capacity of PV and WT, and the cost of utilising the ESaaS.