# Marcus Delgado, Director of Revenue Operations at Pacelane — read of Forecast AI, May 6, 2026

> 9 years in sales ops and RevOps, currently running forecasting and pipeline health for a 75-person Series B SaaS shop in Austin. We have 18 AEs on Salesforce and I own the number every Monday morning.

## How I got here

I was Googling "salesforce quota risk forecasting tool" on a Tuesday at 6:45am during my commute because our CRO called me at 9pm the night before saying she had "no visibility" into Q2. We use Clari and she still says that. I was looking for ammunition, not a replacement, and this showed up on page one organic. I clicked because the meta description had "quota risk" in it, which is exactly the phrase I had typed.

## What I clicked first

The headline: "Sales managers forecast with gut feel. RevOps managers forecast with data." I'm not going to pretend that didn't land. That is my actual argument in every QBR. I screenshotted it. But that's a positioning line, not a product. So I kept reading.

## Where I paused

This sentence: "No setup, no training, no black boxes." They say the model is "trained on 500K+ deals across industries" and also that there's "no training." Those are in direct tension. Does "no training" mean I don't have to configure it, or that it doesn't learn my data? Because those are very different products. A general model trained on 500K heterogeneous deals is not the same as a model that understands that our average sales cycle is 47 days and our enterprise segment closes 40% slower than SMB. That distinction is load-bearing for whether this is useful to me.

## What I distrusted

Three things.

First: "High confidence." That phrase appears once with zero definition. Confidence interval? Model accuracy on holdout data? What percentage of their 30-day-early quota alerts turn out to be right? If they can't put a number on it, I can't sell it to my CRO.

Second: There are no logos, no case studies, no quotes. Not one. A tool claiming to be trained on 500K deals has presumably been used by someone. The absence of any social proof on a page this short reads as either very early or very deliberately withheld.

Third, and I almost missed it: "Built by Wishdeal Studio" in the footer. That tells me this is a studio product, not a dedicated company. That matters. Who do I call when the model starts misbehaving at month-end close? Is there a RevOps person on the other side who has ever done my job, or is this a dev shop that spun up a forecasting wrapper?

## What would convince me

One published accuracy benchmark. Not "high confidence," but: "Our 30-day quota alerts are accurate 78% of the time across 200 accounts, with a false positive rate of 12%." I would actually read a methodology PDF if they linked one. Even a flawed one. The willingness to publish it tells me they built something real.

Also: one customer story from a RevOps person at a company between 50 and 150 employees who switched FROM Clari or Boostup and can say what actually changed. Not "we saw 30% pipeline accuracy improvement," but "here is what we caught that we would have missed."

## What I'd ask in an email reply

1. The model says it's trained on 500K deals across industries. Does it adapt to my specific sales motion over time, or am I always running on the general model? What does "no training" actually mean in practice?

2. What is the measured accuracy on your 30-day quota miss alerts, and how do you track that internally?

3. Who is on the team? Is there a full-time product person who came from RevOps, or is this being maintained by the studio that built it?

## Verdict: on-the-fence

The positioning is the best I've seen on a page like this, and the $299 price point is low enough that I'd expense it without a procurement fight. But I cannot recommend a forecasting tool to my CRO without a single accuracy claim or a single customer name. The page reads like someone who knows the problem well but doesn't yet trust their own product enough to show proof.

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