Here’s the thing that sticks out: almost none of these firms lost out because their competitors were better.

• One had thirty years in business and ten thousand placements, but that got stated nowhere but in an easily overlooked stat box on the homepage.
• Another had been named a top-ten fastest-growing staffing firm in the country, twice, and a two-time “best place to work” winner, and neither fact could be found by a crawler. A human could find it by clicking around the site, but that won’t help the firm win AI search.
• A third had the best blog cadence I’ve seen in this category in a long time: near-daily posts, genuinely useful content, but every single post returned a “JavaScript is required” error to anything that tried to read it without a browser.

There’s a term for this: the citation gap, the distance between what a firm has actually earned (and may have published or promoted, but not in AI-search-friendly ways) and what a search engine or AI system can actually see. And what the bots can see…that’s a lot of the game these days.

Here’s the simplest example of this gap I can think of:

• You’re in a segment like staffing, and your firm has placed more candidates in a certain niche than anyone else in your market, but that number lives in just one paragraph on your About page and is never repeated anywhere else, and never turned into a headline claim.
• A competitor with half your track record writes a single blog post titled “Why We’re the Top Choice for [Your Niche]” and states their smaller number plainly, twice, in the first hundred words.

In this situation, when someone asks AI search who they should hire, it cites the competitor. 😥 

Not because they’re better. Because they closed the citation gap and you didn’t. Every firm in this audit was carrying some version of it. Most of them didn’t know.

The stuff that isn’t hidden, it’s just unreadable

Old SEO punished you for having thin content. Now AI search punishes you for having good content a machine can’t read, which is a different and, frankly, dumber way to lose. It’s why AEO/GEO is suddenly so necessary.

For instance? One firm’s entire site, with real placement numbers, an active blog, a free tool candidates actually used, rendered as a blank title tag to a plain fetch.

Not slow. Not thin. Invisible. The content existed. It just lived entirely in JavaScript that never executed for anything that wasn’t a browser with a human attached. Like an AI crawler.

Another had a bot-management wall (the kind that’s supposed to stop scrapers) sitting in front of every page on the site, homepage included. That firm was publishing nearly every day. Technology recruiting one week, engineering staffing the next, a piece on reducing time-to-hire after that. All of it dated, all of it specific, all of it running headfirst into a wall that said, more or less, come back with a browser.

I have no idea whether that wall is letting through the crawlers that actually matter now, GPTBot, ClaudeBot, PerplexityBot, and neither does the firm, I’ll bet. That’s not a marketing gap. That’s tantamount to a paywall blocking off the best content strategy out of all the sites we audited.

And then there’s an especially galling version of this. We came across a homepage proof section had been built to hold real proof points, numbers, years in business, candidates placed, retention rate, and used a “stat counter” to animate this in a CSS/JS layer. But it was implemented incorrectly, so as far as a crawler can see, there’s just a literal, garbled string of digits on the raw page. Not a zero or a blank but just…garbage.

If your homepage’s headline stat renders as 1out of every0123456789,012345678901234567890123456789 to a bot, then you’ve found a great new way to fail at digital marketing that nobody had to worry about three years ago.

The achievement nobody bothered to move

This one’s almost worse because it’s so easy to fix.

A firm I audited had been ranked one of the fastest-growing staffing companies in the country, two separate years, by the industry’s own analyst group. One of the youngest firms ever to crack that analyst group’s list of the largest staffing firms in the U.S., against a field averaging decades more tenure. That fact lived in exactly one place: a single blog post from a couple years back. Not the homepage. Not the About page. Not anywhere a comparison query would ever find it.

Another firm had the single most citable proof point in this entire audit, a genuinely rare combination of tenure and scale, and it sat in three stat boxes with no supporting page, no case study, nothing extending the claim into something a search engine could actually use.

Here’s the test I’d apply to your own site right now: pick your best, most specific, most verifiable achievement. Not “award-winning.” Not “industry-leading.” The actual number, the actual ranking, the actual third-party name behind it. Now go find it.

If it takes you more than one click from your homepage, an AI system looking for the answer to “who’s the best staffing firm for X” isn’t finding it either, and it’s citing whoever put theirs on the front page instead.

The great blogpost that can’t go it alone

One firm in this batch is, right now, the single cleanest citation win I found across all twelve audits. Ask an AI system for the best recruiting firm in its exact regional niche, and it names this firm first, correctly, quoting the firm’s own language back almost word for word. That citation was earned by one specific, dated, useful blog post about a regional hiring trend.

Then I checked the full posting history. Three posts in one month, then nine months of silence. Until one more recent post showed up. Nothing since, going on five months as of the audit.

A citation built on freshness doesn’t collapse all at once. It erodes, quietly, the same way your competitor’s site inches past yours in a ranking you’re not checking anymore. Whichever competitor’s still publishing when this firm’s win expires gets the next citation instead.

Nobody sends you a notification when that happens. You just stop showing up, and by the time you notice, you’re rebuilding from scratch instead of maintaining a lead you already had.

What fills the silence isn’t neutral

This is the part that should actually worry a marketer in any space, more than any of the above.

When a firm hasn’t published enough to compete for its own name, something still shows up when someone searches it. It’s just not the firm’s choice  as to what that something is.

One firm’s reputation query returned a review site’s rating and a quote calling the experience “the absolute worst… a complete nightmare.” Another firm’s site was down (temporarily, as it turned out, but the AI answering the question didn’t know that), and into the resulting vacuum stepped a multi-year-old government insolvency filing for a related entity, corroborated by employee accounts of a shutdown with unpaid wages.

A third firm, thirty years and ten thousand placements deep, got a “reviews” answer built entirely from complaints about missed paychecks, because nothing the firm had published outweighed it.

None of these firms are really defined by these points. Yet all of them are, right now, being negatively defined to anyone asking an AI system about them. Because nothing more current or more favorable exists to compete with the bad stuff.

There are “paused” (a.k.a. abandoned) housing contruction projects in my son’s area of West Hollywood. After a year or two, they’ve become coyote colonies. What’s the analogy here?

Silence isn’t neutral. It’s a vacancy, and something always moves in. Not necessary a something you like.

A fake testimonial is worse than no testimonial

I want to really squat over this one, because it’s the single most avoidable failure I found, and it’s the kind of thing that takes five minutes to fix, but someone still hasn’t. And until they hire us, I won’t be the one to tell them.

This firm has real, strong, verifiable client logos, the kind of consulting and professional-services names that would make any competitor jealous. Sitting directly underneath that logo wall: four testimonials that were obviously never edited from a website template. Fake names. Fake job titles. Fake companies with names like “Acme Inc” and “WidgetCo.” Quotes about “boosting engagement by 300%” that have nothing to do with recruiting, because they were written for whatever industry the template was originally built for. One name even shows up twice.

Here’s the part that should really sting: even the testimonials on that page that are real carry the tell. The image file for one genuine, named reviewer is still tagged in the code with a different, generic placeholder name. The whole section was built from a template and only half-finished.

I don’t think this firm is dishonest. I think somebody built the site fast, swapped in a few real quotes, and never went back to delete the rest. But intent doesn’t matter to the person, or the AI system, evaluating the page. Once you notice one fake testimonial, you stop trusting the real ones next to it. And once you stop trusting the testimonials, you start wondering about the client logos above them.

A fabricated proof point doesn’t sit quietly next to real proof. It contaminates it.

Having the right content and pointing it at nothing

Not every failure in this audit was a technical one. Some firms had genuinely good, well-shaped content and just never aimed it at the question a buyer actually asks.

One firm has a real FAQ section with real buyer questions (“What is the hidden value of using a staffing agency?”), plus two working interactive tools, a salary calculator and a time-to-hire calculator. This is, on paper, some of the best AEO-shaped content I found in the whole batch. That firm still lost its single most obvious category query, “best technology staffing agency” in its own city, to five competitors with thinner sites.

The content existed. It just wasn’t built into a page that answered that specific question. Having the right ingredients doesn’t do you any good if you can’t get people to see your solution.

What to actually do about it

Every fix described above costs a fraction of what the citation loss is worth.

Delete the fake testimonials. Today. It’s four sentences and a page save, and it’s currently undermining your best real proof points for free.

Check whether your blog’s images have any noindex tags hiding in the theme or plugin settings, especially the page holding your best case studies or press mentions. It’s a simple fix, not a redesign.

Pull your single best, most specific, most verifiable stat off whatever page it’s stranded on and put it somewhere a comparison query would actually find it. Ideally as a sentence, not a counter that might render as blank.

And if you’ve got a blog that used to be good and quietly stopped, the fix isn’t a content strategy overhaul. It’s one more post, on the same topic that already worked, published before the citation you’re still riding on finishes eroding.

None of this requires more budget than most firms already have earmarked for “marketing, eventually.” All or any of it just requires someone to actually go look at what a crawler spots instead of what a human clicking around the site sees. Get them in sync, or else the version of your site that your buyers are reading may not be the one you think is being seen.