When we started building the dispatch logic for encosa's platform, we spent a lot of time modeling what "manual scheduling" actually looks like in practice — not an idealized straw man, but the real operations patterns of commercial facility managers and energy engineers running batteries with weekly or daily schedule updates. Understanding that baseline was essential because it defines the delta our platform has to beat in practice, not just in theory.
Over the 18 months from October 2024 through March 2026, we've accumulated enough operational data — across a range of deployed systems with different configurations, load profiles, and grid connection types — to characterize the performance gap between automated real-time dispatch and manual scheduling with reasonable confidence. This article is our honest read of where the gap is real and where it's smaller than people expect.
Defining the Comparison Fairly
First, we have to be clear about what "manual scheduling" means in this comparison, because the term covers a wide range of sophistication.
At the low end: a battery with a fixed daily schedule programmed once and rarely changed. Charge from 00:00–06:00 (low-tariff period), discharge 07:00–09:00 and 17:00–19:00 (peak demand periods). No market price signal integration. This is genuinely what many C&I batteries run on — especially older installations with limited BMS API capability.
At the high end: an energy manager who checks EPEX Day-Ahead prices each evening, updates the next day's charge/discharge schedule to target the expected price spread, and reviews FCR tender results weekly. This operator is doing roughly what an automated system does with Day-Ahead prices, but without the 15-minute intraday continuous market access, the real-time SOC optimization, or the cross-revenue-stream coordination.
The comparison below uses the high-end manual case as the baseline. We're not claiming victory over an unmanaged fixed schedule — that's too easy. We want to know: what does a well-managed manual operation miss that automated dispatch captures?
Revenue Gap by Category: 18-Month Summary
| Revenue category | Manual (high-end, Day-Ahead based) | Automated (encosa, intraday 15-min) | Gap |
|---|---|---|---|
| FCR availability payments | 88–91% of theoretical max | 93–95% of theoretical max | +3–6% |
| EPEX spot spread capture (Day-Ahead prices) | 62–70% of optimal | 68–74% of optimal | +6–8% |
| EPEX spot — intraday spike capture | 15–25% of available value | 55–65% of available value | +35–45% |
| Negative-price charging events | 22–35% capture rate | 60–70% capture rate | +30–40% |
| Peak shaving (Leistungspreis) | 70–80% of theoretical savings | 78–88% of theoretical savings | +8–12% |
| Cycle count (full eq. cycles/year) | 180–250 | 300–420 | +60–80 cycles/yr |
The FCR gap is the smallest. A well-managed manual system that checks regelleistung.net weekly and maintains proper prequalification achieves most of the FCR revenue. The incremental value from automated FCR management comes from optimizing SOC positioning to maximize available time in the 20–80% band — roughly a 3–6% improvement in effective availability.
The intraday spike capture gap is the largest, and it's structural rather than a matter of diligence. Manual scheduling based on Day-Ahead prices simply cannot capture events that emerge in the intraday market 1–4 hours before delivery. No amount of more careful manual schedule management bridges that gap — it requires real-time market access and automated execution.
Where Manual Scheduling Performs Well
We're not saying manual scheduling is a bad approach to battery operation. There are real scenarios where its performance is close to automated dispatch, and understanding these limits the case for automated dispatch to where it genuinely applies.
Low-volatility market periods
In quarters where EPEX spreads are narrow and predictable — Q4 2024 was a reasonable example — the Day-Ahead price ladder is a good proxy for intraday prices. Intraday deviations from Day-Ahead are small, intraday spikes are rare, and a manual schedule built from Day-Ahead prices performs close to optimally. The incremental value from continuous intraday monitoring in those periods is 8–12% above manual — meaningful for a high-utilization asset, but not the 3–4× gap that appears in high-volatility events.
FCR-only operation
If a facility's primary goal is pure FCR availability revenue — no spot trading, simplified operation — manual management with weekly regelleistung.net tender submission and a fixed SOC band can capture 85–92% of theoretical FCR revenue. Automated dispatch adds 3–8% in this scenario. For an operator who genuinely doesn't want to trade intraday or manage SOC dynamically, the fixed-cost overhead of an automated dispatch platform needs to be justified against this smaller delta.
Highly predictable load profiles
Facilities with very stable, predictable load patterns — a cold storage warehouse with consistent thermal load, a factory with a fixed shift schedule and minimal load variation — allow manual operators to set a schedule that remains near-optimal for weeks at a time. The advantage of automated re-optimization shrinks when re-optimization doesn't change the answer much day to day.
Where Automated Dispatch Structurally Outperforms
Intraday spike events
This is the clearest structural advantage. Consider the Q3 2025 heatwave events described in our market review article. EPEX intraday 15-minute products for the 15:00–17:00 window on July 16, 2025 cleared above €220/MWh — but the Day-Ahead price for the same window was €118/MWh. The incremental €100+/MWh appeared only in the intraday continuous market, with final prices confirmed at 13:45 for 15:00 delivery.
A manual operator who set their schedule at 22:00 the previous evening (from Day-Ahead data) had a charge window planned for the morning and a sell window planned at 15:00–17:00 at expected €118/MWh. They captured that. They didn't capture the additional €100+/MWh intraday premium because there was no mechanism to observe, decide, and execute in real time without being physically present at a trading terminal at 13:45.
Automated dispatch observed the intraday order book rising from 11:00 onward, confirmed the 15-minute product prices exceeding the profitability threshold at 13:10, and pre-positioned SOC for maximum discharge availability at 15:00. Total captured value on that day's cycle: roughly 2.3× what the Day-Ahead-based manual schedule captured.
Overnight and weekend events
Manual scheduling requires human attention. At 02:30 on a Sunday morning, when EPEX intraday negative prices appear due to a wind surge that wasn't in the weekend forecast, a manual system is idle. An automated dispatch system is reading the 15-minute product prices and issuing charge commands. The same applies on bank holidays, during summer vacation weeks, and any time the energy manager isn't actively monitoring the market.
The annualized impact of overnight and weekend events: in our data, approximately 22–28% of high-value dispatch events (intraday spreads exceeding €80/MWh) occur between 22:00 and 07:00 or on Saturdays and Sundays. That's a significant share of annual revenue opportunity that manual scheduling cannot systematically capture.
Multi-constraint SOC optimization
A battery stacking FCR + spot + peak shaving simultaneously faces a continuous SOC constraint management problem. FCR wants 40–60% SOC. Spot trading wants enough headroom to charge low and discharge high. Peak shaving wants discharge capacity available during the 10:00–12:00 and 17:00–19:00 windows. These constraints interact, and the optimal SOC position changes every 15 minutes as prices and load forecasts update.
Manual scheduling resolves this conflict by simplifying it: pick one primary use case and treat the others as secondary. In practice, most manual operators default to FCR as primary, add a fixed spot trading window on top, and rely on a passive peak-shaving residual. This simplification leaves money on the table whenever the constraints allow a better configuration — and in a typical day, they allow it for several hours.
The value of real-time multi-constraint SOC optimization in the 18-month data: approximately 12–18% additional revenue compared to a well-managed fixed-priority manual schedule, averaged across all market conditions. In high-volatility periods, the premium is higher.
The Honest Counterpoint
Automated dispatch is not frictionless. There are three categories of failure that reduce the real-world gap from the theoretical maximum.
Communication reliability: Automated dispatch depends on continuous reliable data from the BMS, the smart meter, and the market data feed. A BMS that goes offline for 2 hours due to a network issue, a smart meter that reports stale SOC data, or a market data feed latency spike — these cause dispatch errors that a human operator watching a screen wouldn't make. In 18 months, we've seen these issues contribute to missed dispatch events equivalent to roughly 2–4% of annual revenue potential. That's real, and it's an engineering problem we're continuously working on.
Over-trading risk: Automated systems can execute more cycles than a manually operated battery would. More cycles means more cycle wear on the battery. If the dispatch model slightly overestimates the revenue value of marginal cycles, it will run more cycles than is economically optimal on a total lifecycle basis. We calibrate our cycle cost model against LFP degradation curves, but the calibration is imperfect and the battery's true marginal degradation cost per cycle is uncertain. Manual operators, paradoxically, may preserve battery life through conservative scheduling — the value of which shows up in year 7–10 of the battery's lifecycle.
Regulatory instruction compliance: Automated systems operating at scale must handle Redispatch 2.0 and TSO instructions correctly and immediately. A missed or delayed response to a grid operator instruction carries financial and regulatory penalties. The compliance burden is manageable in an automated system but requires deliberate engineering — it's not a free parameter.
The Actual Revenue Differential at System Scale
Putting all the above together for a 200 kWh / 100 kW system in Bavaria, annualized over 18 months and normalized:
| Dispatch approach | Estimated annual revenue (200 kWh, Bavaria) | vs manual high-end baseline |
|---|---|---|
| Manual — fixed schedule (low end) | €48,000–€55,000 | — |
| Manual — Day-Ahead managed (high end) | €68,000–€78,000 | baseline |
| Automated — Day-Ahead only, no intraday | €74,000–€84,000 | +8–12% |
| Automated — full intraday 15-min + multi-constraint SOC | €88,000–€105,000 | +25–35% |
The 25–35% premium over a well-managed manual operation is our honest estimate of the sustainable automated dispatch advantage in the current German market. In high-volatility quarters like Q3 2025, the premium was larger — closer to 40–50% for some installations. In low-volatility periods, it compresses toward 10–15%.
Over a 10-year asset lifecycle, a 25–35% annual revenue premium on a 200 kWh system translates to €200,000–€300,000 in additional cumulative revenue net of platform fees. That's a significant fraction of the original CAPEX recovered as incremental value from the dispatch layer alone.
What This Means for the Deployment Decision
The question isn't whether automated dispatch beats manual — it does, under almost all market conditions — but whether the automated system you're evaluating actually delivers on the intraday execution capability that produces the largest part of the gap. That means confirming:
- Does it access EPEX Intraday continuous 15-minute products, or only Day-Ahead?
- Does it perform multi-constraint SOC optimization across FCR, spot, and peak-shaving simultaneously?
- Does it operate continuously overnight and on weekends without human intervention?
- What is the measured uptime and data latency for BMS communication?
- How does it handle Redispatch 2.0 instructions and re-optimize the remaining day?
These five questions separate platforms that deliver the full automated dispatch advantage from those that are essentially automated Day-Ahead scheduling with a nicer interface. The difference between those two things, in annual revenue terms, is the story of these 18 months of data.