# Marcus Thielen, Director of Asset Management at Clearfield Energy Partners — read of renewable-energy-analytics, June 5, 2026

> 14 years in renewable O&M and dispatch, currently managing a 38-asset portfolio across solar and wind in MISO and WECC. Coaches U10 soccer Saturdays. Reads vendor pitches on his lunch break.

## How I got here

LinkedIn ad, which already had me slightly suspicious. The targeting was broad enough that I figure everyone who's ever written "renewable" in a headline got it. I clicked because the headline wasn't trying to be clever and the subhead actually named my role: "utilities, renewables operators, and energy consultants." I've closed worse tabs faster.

## What I clicked first

The forecasting accuracy claim in the hero pulled me in: "±5.2% MAPE on 24-hour horizons." That's a real number with a real unit. I've sat through probably 40 vendor demos where they say "industry-leading accuracy" and stop there. Whoever wrote this copy knows that MAPE means something to my team. I kept reading.

The SCADA list in the FAQ also got my attention: "OSIsoft PI" is on there. That's not a name you drop if you don't know the space. PI is everywhere in our ops center.

## Where I paused

The FAQ says "4.8-5.8% MAPE" and the feature section says "±5.2% MAPE." Those are not the same number presented two different ways. That's two different numbers. Either the FAQ range is narrower than the hero, or someone edited one and forgot to update the other. For a product making accuracy its main claim, that's a sloppy tell. I actually highlighted both and pasted them into a note to check later.

## What I distrusted

Pretty much everything after the FAQ section. I scrolled past the standard pricing table -- fine, $899/month for fleet, we'd spend that on a single turbine blade inspection -- and then I hit this:

"Honest disclosure: we don't have live customers on this idea yet."

Hold on. "This idea." Not "this product." Not "this platform." Idea.

Then: "We shipped the strategy package; you ship the customer conversations."

And then the adoptability score. The Fermi math. "1 in 11 meaningful-success odds." The pricing tiers for buying the dossier. "$5 to unlock ICP, MVP scope, first 7 build tasks."

This page is not a product. It is a pitch for me to go build a product. The top third is a mockup of what the product could look like, written convincingly enough that I spent eight minutes reading it before I realized there's nothing here to actually buy except a $99 code starter kit and a GTM outline.

The MAPE numbers, the FERC compliance language, the SOC2 certification, the SCADA connectors -- all of it is hypothetical. It's designed to show a potential founder what the pitch could look like. I'm not the buyer. I'm the market research.

## What would convince me

This question is almost beside the point, because the product doesn't exist. But if someone built this and came back:

I'd want to see one real dispatch center, with a name I can look up, that ran 12 months of day-ahead forecasts through this model and let me compare the MAPE against their existing process. Not a backtest on historical data. Live forecasts. I've seen backtests that show great numbers fall apart in real dispatch conditions because curtailment events, grid emergencies, and equipment faults don't show up cleanly in historical actuals.

The "30-40% better than industry benchmarks" claim needs a citation. What benchmark? EPRI? NREL? One of the ISO studies? That number means nothing without a source I can look up.

## What I'd ask in an email reply

1. The MAPE on the homepage says ±5.2% and the FAQ says 4.8-5.8%. Which is it, and is that across all asset types or is solar separate from wind?

2. When you say "retraining on your historical actuals" -- what's the minimum data volume before retraining actually improves accuracy over the base model? We have sites under two years old.

3. What does OSIsoft PI integration actually mean in practice? Read-only historian pull? Real-time tag subscription? We had a vendor last year who said "full PI integration" and meant they could ingest a CSV export.

## Verdict: dismissive

The top of the page is genuinely well-written for the audience. Whoever put together the forecasting section knows the vocabulary. But it's a studio mockup selling me a business idea, not a product I can evaluate. I don't have time to become the founder of this.

---
*Memo by skeptic persona, generated 2026-06-05. Studio breaks own self-grading loop.*