How Puntt AI Cut Indeed's Marketing Review From 3 Days to 2 Hours

This case study shows how Indeed's marketing review team used Puntt's knowledge-backed AI to cut content review from three days to under two hours, and sustain it for eight straight quarters. We sat down with Jennifer Waits, the Senior Manager of the Marketing Review team,  at Indeed's Austin, Texas headquarters to hear how it happened.

What Puntt AI delivered

• Triage time BEFORE Puntt AI: Up to 3 days
• Triage time AFTER Puntt AI: Under 2 hours (sustained for 8 consecutive quarters)
• Overall review time: 4.23 days, reduced to 1.83 days
• Accuracy: 89% → 93%, held quarter over quarter
• Legal escalations: Down 21%
• Reviewer time saved: ~4 hours per reviewer, per week - now used for strategic work
• Value proven in 2 weeks; every quarter since rollout has beaten the last

‍

The problem Puntt was brought in to solve

Indeed's Marv (marketing review) team is the gate every piece of external content passes through — marketing, sales, CS, PR, government relations — across ten markets, each with distinct claims laws and regulations. Four reviewers handled roughly 400 tickets a quarter, and onboarding a single reviewer took over a year.

The hardest, highest-stakes step was triage according to Waits:

"Triage was the most comprehensive, the most complex, and where the stakes were the highest. If you failed to send something to legal that needed to go, that's a problem. If you miss something and leave guidance out, that's a problem."

Then generative AI multiplied inbound volume tenfold, and the manual model broke. The team had exhausted every lever — new intake processes, org-wide training — and concluded that only one kind of solution could keep pace:

"It took something driven by AI to match the power of AI."

‍

How Puntt automated the compliance review workflow

1. Puntt integrated directly with Indeed's Approved Data Hub and erased the most manual part of the job.

The single biggest time sink in triage was hunting down data claims and citations against Indeed's live database of cleared product data, by hand, dozens of times a day. Puntt connected to that hub seamlessly, and the bottleneck collapsed almost immediately, said Waits:

"The tool was able to integrate with the Approved Data Hub seamlessly, and immediately we saw gains by multiple hours. The triage process that could take up to three days was reduced dramatically to under two hours. If we never took puntt anywhere else and just let it ride, we've already seen such a return on investment — simply because we don't have to spend the time manually checking data anymore."

‍

2. Puntt AI was built to be trained like a team member, not just queried like a chatbot.

Because Puntt could be taught to recognize content types and their components, Waits said Marv shifted from training a dozen humans to teaching one system that scales like a hundred.

"Instead of training 10 or 12 employees, the challenge became: how do you teach this one AI agent to take on the work of 100? While we were teaching ourselves how to use AI, we were learning how to teach puntt what we needed it to do."

‍

3. Puntt delivered consistently. And fast.

Waits was quickly impressed:

"Within two weeks we knew we had something that was not only working effectively, but consistently working — which is a very big deal with technology. There hasn't been a quarter since rollout that we haven't surpassed the previous quarter's goal on accuracy, turnaround, and cycle time — and that's at scale, with this explosion of AI-generated content."

‍

Why not a general-purpose AI like ChatGPT instead of Puntt?

This is the heart of Wait's case for the product. She's tried general AI tools for review, and draws a hard line between them and Puntt:

"Puntt is a knowledge-backed system — built on guidelines, regulations, laws, and structured data, with a ton of oversight, provable and improvable. The others can feel editorial, opinion-based — a 'perhaps' — versus concrete answers grounded in source data. For the integrity of the things I need to pay attention to, that's a huge difference."

On reviewing content through tools like GPT:

"Without eyes into the build of the knowledge hub behind it, it feels like an opinion base — editorial, a 'perhaps.' That's the opposite of what Puntt does, which is concrete, based on source data and factual information."

And critically, she credits puntt's accuracy gains to the fact that it's a knowledge-grounded system humans can steer — not a black box:

"Junk in, junk out — the accuracy rate is on you. You fuel the tool with the prompts, led by curiosity and strategy. Puntt gives you a system worth fueling."

‍

Real Results for Marketing Review

The downstream effects matter as much as the headline speed. By resolving more issues upstream, Puntt cut legal escalations by 21% and freed the team to hold 38 dedicated work sessions with their marketing-legal partners in 2025 — proactively clearing compliance, privacy, and claims edge cases instead of routing tickets. Each reviewer now spends about four hours a week on strategic consultations that help marketers plan better and build review into their process earlier.

Waits — who openly rejects "vanity data" and "cook-the-books data" — stands behind every number:

"Sometimes our data is just so dreamy that people don't believe us, because it's so good."

‍

What made the rollout work: the Puntt partnership

A recurring theme is that Puntt's team moved at the speed of Indeed's business and stayed in the trenches with Marv.

"The expertise from the Puntt team, the guidance, the accessibility — it made it actual. We come to the table with ideas, we bounce them off your team, you show us how to think about it from Puntt's perspective, and then we go figure out how to teach the tool. You've been there with us the whole time."

The takeaway for teams evaluating Puntt

Wait's advice doubles as the case for the product: find your most painful, most manual, highest-stakes process — for Marv it was triage and citation-checking — and let a purpose-built, knowledge-backed system take it on.

"Puntt freed us to stop being paper-pushers. You're a thinker, not a reviewer — and a tool like this lets you empower your team to be thinkers, too."

Check out the full interview below: