Spotlight.ai in Today US: Every Box Was Green. The Deal Was Dead.
A new contributed article in Today US takes on the AI build most revenue teams are running right now, and explains why a qualification record full of green boxes keeps producing deals that close lost.
BOSTON. Lolita Trachtengerts, VP GTM Ops & Growth at Spotlight.ai, has published a contributed article in Today US titled "Every Box Was Green. The Deal Was Dead." The piece argues that AI has made it trivial to fill in a qualification record and done nothing to prove the deal behind it.
It opens on a deal that closed lost. Go back sixty days, before anyone knew it was in trouble, and read the qualification fields. Economic buyer identified. Pain identified. Champion confirmed. Every box green. The deal still died. The record was complete, and none of it was true.
Three ways a filled field misleads you
The article names three structural problems with letting a model read one transcript and fill the boxes.
Evidence and opinion share one field. A buyer saying their renewal starts in March and a rep believing the buyer has budget end up stored the same way, so the record cannot tell what was heard from what was assumed.
One call stands in for the whole deal. A model reading a single transcript cannot tell a buyer's passing comment from a commitment repeated across three meetings, because it only ever saw the one.
Nothing re-checks the answer. A field marked identified in March stays identified through a reorg on the buyer's side and a champion who has gone quiet, long after it stopped being true.
The problem is the question
The core of the piece is that the bad field comes from the question. Hand a conversation to a general model and ask which MEDDPICC box it belongs in, and the model will always return an answer, because it has no way to return nothing. When the evidence is thin, it produces a plausible sentence where a gap should be, and that sentence reads exactly like a real one.
Trachtengerts then describes the design choice behind Spotlight.ai. Every question the engine asks has a closed set of possible answers defined before a model is called: yes or no with a probability, a choice from a known list, or a position on a defined scale. Qualification status, the deal score, gaps and the forecast are then assembled from those answers in code. The conclusions are computed rather than written.
What changes for the forecast
The article lists three effects. The same deal returns the same answer on Monday and Friday, so a score only moves when the deal moved. A question the evidence cannot answer surfaces as a visible gap with low confidence attached, instead of a confident guess. And because nothing is written at the decision layer, the team can interrogate the pipeline as often as it likes without the bill growing with its curiosity.
How Spotlight.ai separates evidence from opinion in the product: Validating AI results in Spotlight.ai (Spotlight.ai Support)
The results behind the argument
The piece grounds the case in two customers. Tulip ran evidence-based qualification across a 65-person sales team and moved conversion from 6.3% to 15% over twelve months, a 2.4x improvement. Sysdig saw a 3.8x pipeline conversion over the same period. In both, the change was not more fields filled, it was fields that could be trusted.
"A field being populated doesn't mean the deal is qualified," Trachtengerts concludes. "We automated the filling in. Nobody automated the proving."
Read the full article
Request a demo
Spotlight.ai qualifies deals from evidence across every call, email and meeting, keeps evidence and opinion separate, and flags what it doesn't know as a gap on the deal rather than filling it with a guess.
About Spotlight.ai
Spotlight.ai is the autonomous deal execution company. Headquartered in Boston, it replaces manual qualification, inspection and forecasting with an AI agent squad that listens to every buyer interaction, qualifies deals against the customer's own playbook, generates customer-facing assets and produces evidence-based, bottom-up forecasts. The platform has autonomously qualified and executed more than $8B of pipeline and is SOC 2 Type 2 certified.
About Lolita Trachtengerts
Lolita Trachtengerts is VP GTM Ops & Growth at Spotlight.ai, where she leads go-to-market operations and growth. She co-leads the Boston chapter of RevGenius and writes regularly on RevOps, qualification and how AI is changing the work of revenue teams.
FAQs About AI-Filled MEDDPICC and Evidence-Based Qualification
What does "every box was green" mean?
It describes a qualification record that is fully populated and still wrong. The fields all read yes, often filled by a model reading a single call, yet nothing confirms the answers were backed by evidence or are still true, so a deal can look fully qualified right up until it closes lost.
Why isn't a transcript-to-ChatGPT-to-Salesforce build enough?
It treats one call as the whole deal, stores evidence and opinion in the same field, and never revisits an answer once it is written. It also asks the model an open-ended question, which always returns an answer even when the evidence is thin.
How does Spotlight.ai avoid hallucinated qualification fields?
Every question the engine asks has a closed set of possible answers defined in advance, such as yes or no with a probability or a choice from a known list. Conclusions like qualification status and the deal score are computed from those answers in code rather than written by a model.
What happens when there isn't enough evidence to answer a MEDDPICC question?
The answer comes back with low confidence and appears on the deal as a visible gap, so managers see what is unknown instead of a confident guess that fills the box.
Where was the article published?
The article was published in Today US as a contributed piece by Lolita Trachtengerts, VP GTM Ops & Growth at Spotlight.ai.
Comments