Five questions to ask before you buy multi-brand solar monitoring

Most renewable energy portfolios don't start multi-brand. They become it. One EPC installs Huawei, the next installs SMA, a battery project brings in Victron, and three years later the operations team logs into four vendor portals every morning, each with its own KPIs and its own alarms.

At that point a portfolio monitoring platform looks obvious. The hard part is telling a platform that really unifies your fleet from one that shows a nice dashboard in the demo and falls apart on your data. These five questions do that. Ask them before the demo, and ask for the answers in writing.

1. Which of my brands connect by cloud API, and which need hardware on site?

This is the question that decides your rollout cost. Most current inverter brands offer a cloud API: Huawei, SMA, Sungrow, SolarEdge, Fronius, GoodWe and others. Connecting those is a configuration job, usually done in days. Older inverters, gensets, some meters and anything behind a site firewall often need a datalogger on site.

Send the vendor your device list by brand and model, and ask for a table back: API or logger, typical setup time, and one-off cost per site. A platform that says "we integrate with everything" without that table hasn't looked at your fleet yet.

2. How are KPIs calculated across brands?

Every inverter portal calculates performance ratio, availability and yield its own way. Some use plane-of-array irradiance, some use satellite data, some leave out curtailment and some don't. Put three brands side by side and you're comparing three different definitions.

Ask which standard the platform uses (IEC 61724 is the usual reference for PR), where irradiance comes from when a site has no sensor, and whether the method is the same for every brand. If KPIs are only passed through from vendor portals, your portfolio ranking is noise.

3. What happens when a vendor API goes down?

It will. Vendor clouds have outages, rate limits and silent changes to their APIs. The questions that matter: how fast does the platform notice missing data, does it backfill when the API comes back, and is backfilled or estimated data marked as such?

The last point matters most if you bill from this data. An invoice built on estimated values that nobody flagged is a dispute waiting to happen.

4. Can alert rules be set per site, per group and per portfolio?

One threshold across 200 sites gives you one of two outcomes: so many alerts that people stop reading them, or thresholds so loose that real faults slip through. Rooftop sites with shading, ground-mount sites and hybrid sites with gensets need different rules.

Ask to see how rules are set at each level, how vendor alarms are filtered or grouped, and how alerts reach people: email, Microsoft Teams, Slack, webhooks into your ticketing tool. Then ask how many alerts a typical site produces per week. The number tells you more than the feature list.

5. Can you get your data out?

Your monitoring data will outlive any one tool. Ask for a documented API, bulk exports in CSV or Parquet, and whether there are per-seat or per-call fees on either. A newer question worth adding: is there an MCP server, so AI assistants such as Claude or Copilot can query your portfolio directly, with the same permissions as your team?

If getting data out is hard or expensive, switching later will be too.

The short version

You're buying three things: coverage of every brand you run, one consistent way of measuring them, and freedom to take your data elsewhere. The five answers above tell you whether you'll get all three.

AMMP OS is built around exactly these questions. It connects to 30+ vendor platforms by cloud API or the AMMP Edge datalogger, calculates KPIs the same way across brands, and gives you your data through a Data API and an MCP server.

Questions about your own portfolio? Write to sales@ammp.io.