Solutions Spotlight on SparcStart

Chris Hoyt: Jerry, what are we doing today?

Gerry Crispin: I have no idea. No—today we’re talking about features and their benefits to our clients.

Chris Hoyt: I’m excited. We do these spotlights on occasion when we see something really cool bubbling up in the space, and we want to shine a light on the work that’s being done.

Sometimes we do this as a teaser—something a little early, not quite out to market yet. Other times we do it because it’s an early release or it’s been around for a while and maybe you missed it.

I’m excited about this one coming out of Maury’s shop over at SparcStart because this is pretty new. I think it’s a really cool take on a bigger challenge than most folks realize they’ve got in front of them.

Welcome to the spotlight.

Maury Hanigan: Chris, Jerry.

Chris Hoyt: For those who haven’t had the pleasure of meeting you—maybe you live under a rock and don’t know what SparcStart is—do you want to give an overview of you and your organization before we jump in?

Maury Hanigan: Sure. I’m Maury Hannigan, CEO of SparcStart. We’re a 14-year-old recruitment marketing platform.

At our core, we help employers get their story out about who they are, why candidates want to work there, and why they should apply. The goal is to attract people who really understand the positions and apply for the right reasons.

Chris Hoyt: I love it. Now tell us—you’ve got something called Illuminate. Do you want to give us an overview? I think Jerry and I have plenty of questions.

Maury Hanigan: Great.

Illuminate is a new product. It’s a little bit of a departure from what we’ve done before, and a little bit not. We’ve always focused on helping employers communicate their story in a way candidates can actually consume.

We lean heavily on video because that’s how people take in information these days. About 84% of internet traffic is video.

But the real focus is communicating and connecting with candidates. In that sense, Illuminate is a direct extension of everything we’ve done since our inception. The difference is that it’s AI-focused.

One thing people realize at some level—but maybe haven’t fully focused on—is that candidate behavior has changed because of AI. Even before they get to job boards or career sites, they’re sitting down with ChatGPT, Claude, Gemini, or other tools and asking questions that guide their job search.

Right now, employers don’t have a voice in those answers.

Employers often think, “All the information is on our career site.” What they don’t realize is that large language models don’t treat career sites as objective sources of information about organizational culture. They view them as advertising sites, so they skip right past them.

We’ve done hundreds of searches about companies—asking things like, “What’s it like to work here?”—and then looked at the citations. Not once have we seen a career site cited.

So even if the information is there, LLMs skip over it.

The problem is that employers don’t have a voice in the answers. LLMs are pulling information from places like Glassdoor—the great meeting place of disgruntled employees. Even if you’ve kept your employee pages updated on sites like Indeed, the odds of that content being surfaced aren’t great.

The models are finding information they consider credible and objective, and they’re not using employer-supplied information. So whatever they bring back is missing the employer’s perspective.

That’s the problem we can solve.

Chris Hoyt: This is interesting. When you first brought this up, I had to go test it.

I asked questions like “What’s it like to work at ThoughtWorks?” “What’s it like to work at Ford?” and similar things. I was a little surprised to see the ever-trusted platform of Reddit showing up as a source before Ford’s own career site or employer content.

Maury Hanigan: Exactly. Did you even find the career site?

Chris Hoyt: Occasionally, but it was buried. I had to dig for it.

Maury Hanigan: Right. If you ask about jobs—for example, “Who has engineering jobs in electric vehicles?”—then the models might look for job listings.

But if you ask about culture or what it’s like to work somewhere, they go to other places first.

Chris Hoyt: Jerry, does that surprise you?

Gerry Crispin: Yeah, it really does.

Especially when they’re pulling from places like Reddit and other edgy areas without any sort of signal to the user that the information might be sketchy.

I also wonder to what degree sources like Wikipedia might come into play. I would think those should be considered viable sources and used when relevant.

Maury Hanigan: They will go to news sites because those are seen as objective.

But think of it this way: if you ask how good a Ford car is, the model isn’t going to Ford’s marketing site to get a review of Ford cars. That’s not credible information. It’s advertising.

AI also looks at context. If something reads like marketing language, the model can detect that very quickly. It strongly prefers natural, spoken language.

One thing we’re doing is taking employer input and converting it back into natural language.

For example, many organizations have EVPs with carefully crafted messaging—three pillars and very concise phrasing. That works for marketing, but it often doesn’t reflect how people actually speak.

Healthcare organizations might say things like “patient-centered” or “patient-centric care,” as if everyone understands that phrase. But nobody actually talks that way. People say things like, “We care about our patients’ health, safety, and comfort.”

That’s natural language.

As soon as LLMs see phrases like “patient-centric care,” they recognize it as marketing copy and skip it.

Chris Hoyt: Do you think the best way for employers to see how they’re doing is just to ask the AI themselves? Should they go into Claude or ChatGPT and ask questions? Or should they take the results with a grain of salt?

Maury Hanigan: They should ask—but from someone else’s computer.

Seriously. Use a spouse’s computer, a friend’s computer, a neighbor’s computer—anything not associated with you.

LLMs are designed to keep you engaged and provide flattering responses. If I sit down at my own computer and ask about SparcStart, the answer is incredibly positive. Apparently SparcStart is an amazing organization, and the CEO is fantastic.

But if I sit down at my husband’s computer and ask about SparcStart, the answer is still positive but noticeably different.

If you regularly use an LLM for work, it already knows where you work. If you’re in employer branding, it might simply feed your own employer brand messaging back to you.

If you want to see what candidates see, you need to use a neutral device.

Chris Hoyt: You shared a really interesting story with me about a family friend who was using AI to plan their job search. Can you talk about that?

Maury Hanigan: Sure.

This was actually one of my son’s friends—an engineer with two kids. He was thinking about moving out of Los Angeles in a couple of years so his kids could have a more outdoors-oriented lifestyle.

He sat down with an AI tool and typed in what kind of engineering job he wanted and the industry he was interested in.

But then he kept going.

He added that his wife is a professional and that he wanted at least three employers within commuting distance that could offer her a meaningful career. He wanted to be 20 minutes from mountain biking trails, within an hour of a major airport, and in an area where home prices fell within a certain range. He also wanted good schools.

He even asked about things like professional sports teams and local outdoor activities.

This wasn’t just about a job. It was about the entire lifestyle around the job.

Part of what we’re doing is structuring data around the kinds of questions candidates are actually asking. We maintain query libraries by industry and job function because nurses ask different questions than software developers.

For example, one of the top questions nurses ask is, “What’s the patient-to-staff ratio?” That information almost never appears on career sites.

If you want your data to show up in answers, it has to be structured properly. It needs to be machine-readable, tagged correctly, and placed in a credible location.

That’s not traditional SEO. It’s more about knowledge graphs—how AI understands relationships between pieces of information.

And employers often struggle with this because they typically only control websites with their own name in the URL. Publishing credible third-party information at scale is difficult for them.

Chris Hoyt: I remember a time when everything was about keyword optimization, backlinks, and link networks.

This feels like the next evolution—not SEO, but AI optimization.

Can you explain why SparcStart is positioned to solve this problem with Illuminate?

Maury Hanigan: There are a few reasons.

First, we have the technology. We’ve spent 14 years working with video and making it discoverable.

LLMs can’t actually read video. They need transcripts—separate files they can parse. They’re not going to process thousands of video frames. We’ve already built the infrastructure around that.

Second, we’ve always focused on candidates and telling them what they want to know. For example, we never use teleprompters in our video recordings. As soon as you script people, you lose natural language.

Third, we already publish on a credible third-party site.

For 14 years we’ve published hiring manager videos—real employees talking naturally about their jobs. LLMs evaluate things like the age of a site, its reputation score, and how long it’s been publishing.

That credibility matters.

We also structure the data properly with schema so AI understands what the information represents and how it connects to other information. It’s more like a network of relationships than a list of keywords.

Gerry Crispin: Are you now able to measure whether LLMs are actually reading and using this information?

Maury Hanigan: Yes.

We start with baseline data. Clients fill out a simple questionnaire about what they want to be known for and who their competitors for talent are.

Then we run queries across the four major LLMs and generate a baseline score—looking at things like frequency and impressions.

Every 90 days we measure again to see what’s changing.

Launching a product without promises is unusual, but that’s where we are. We can’t promise you’ll show up first. We can’t promise we’ll overwhelm all the existing information online.

What we can promise is that we’ll make your voice discoverable so it appears in the answer set.

There’s also reality involved. If you’re an investment bank and want to be known for a warm, cuddly culture, there’s a lot of existing data suggesting otherwise. We can’t erase that.

But right now, employers don’t have a voice in the conversation at all. With the right structured data on credible sites, they can at least participate in the answer.

And early movers will have an advantage.

Chris Hoyt: So if I’m that warm and cuddly finance institution and I want to get started, what should I be thinking about first?

Maury Hanigan: Start with internal education.

A lot of organizations are focused on things like programmatic advertising and job postings. They haven’t realized that the job search is starting before candidates ever reach those channels.

The number of candidates starting their search with AI is growing exponentially.

Everyone is using AI now—even people who say they don’t. If you use Google, you’re using AI because Google has already integrated it.

At the same time, employers are dealing with massive numbers of applications due to auto-apply and one-click apply tools.

The way to regain efficiency is to increase the percentage of candidates who are genuinely interested, qualified, and available.

Right now, many candidates simply click “apply” on dozens of jobs and let employers figure out whether they’re a fit.

That’s unsustainable.

You need candidates who understand the organization and genuinely want to work there. When that happens, they invest effort in the process and respond when contacted.

That’s when recruiting efficiency improves.

Chris Hoyt: I’d challenge anyone who says they still rely solely on Google. If you try to Google something now, the top of the page is mostly ads.

Even my kids are using AI tools as their search engine.

Maury, I think you’re absolutely right.

Maury Hanigan: Even the summary paragraph at the top of Google results is AI-generated now.

So even if you type into Google, you’re still interacting with AI.

Chris Hoyt: Maury, where should people go if they want to learn more or explore this for their organization?

Maury Hanigan: They can visit sparkstart.com. There’s a page specifically about Illuminate at sparkstart.com/illuminate.

We’re happy to have conversations because many people are just starting to wrap their heads around this.

One thing I worry about is organizations thinking they can simply improve their career site and everything will be fine. That’s probably not going to work.

I often use this analogy: if you’re looking for artisanal cheese from northern Italy and you’re driving past a Kroger, you probably wouldn’t go into the grocery store expecting to find it.

The career site is the wrong place for the kind of candid, credible information LLMs are looking for.

AI tools are searching for trusted third-party sources.

Our goal is to make this turnkey and easy for employers. We handle the structure and the publishing.

And honestly, I’m always happy just to talk about it and help educate people.

Chris Hoyt: If you’d told me six months ago that this was the direction things would go, I probably would have said, “Optimize the site, rebuild your keyword structure, focus on SEO.”

But in the last six months, everything has changed.

Bringing in a partner who specializes in this—rather than trying to figure it out internally in TA, operations, or marketing—seems like a smart move.

Maury, this is exciting. It’s new, it’s cutting edge, and that’s exactly why you’re in the spotlight.

Maury Hanigan: Wonderful. Thank you. I love it.