82% of Clients Expect AI-Powered Service. Only 1 in 4 Firms That Built One See a Return.

Eighty-two percent of accounting firm leaders in a BILL/NewtonX survey of more than 200 firms say AI has raised what their clients expect from them – faster turnaround, sharper advisory input, tighter data privacy, better compliance coverage. Only 29% of those firms actually set a goal to build a new AI-powered service line in response. And among that smaller group who tried, only 25% – one in four – reported a substantial return. BILL's own framing of that last number is the part worth sitting with: it's the highest "significant impact" rate of any AI goal measured anywhere in the survey. The best-performing AI initiative in the whole study still only paid off for a quarter of the firms that attempted it.

That's a specific, well-sourced chain of numbers about accounting firms, and I want to be precise about what it does and doesn't tell a boutique strategy or competitive-intelligence advisor reading this. It isn't a study of your segment. But the shape of the gap it describes – expectation running well ahead of delivered value – shows up everywhere else researchers have looked for it recently, which is exactly why I think it generalizes as a pattern, even though the accounting numbers themselves don't.

The pressure clients apply isn't one thing

BILL's survey breaks the 82% down into what's actually driving it, and the breakdown matters more than the headline number. Speed of service tops the list at 79%, followed by advisory expectations at 67%, data privacy at 66%, cybersecurity at 65%, compliance at 63%, and transparency at 51%. That's not "clients want an AI chatbot." It's clients applying pressure on nearly every dimension of the relationship at once – faster, smarter, safer, more compliant, more transparent – because AI has made all of those things feel more achievable in the abstract, whether or not any specific firm has actually built the capability to deliver on them.

Notice how few of those six pressures are actually about AI directly. Speed, privacy, cybersecurity, compliance, and transparency were expectations clients had before any of this – AI just recalibrated what "acceptable" looks like on each one, upward, all at once. That's a subtler and harder problem than "add an AI feature." It means the bar moved on dimensions you were already being judged on, not just on a new dimension that didn't exist before.

The gap shows up everywhere else too

If this were only an accounting-industry quirk, I'd be more cautious about extending it. It isn't. Deloitte's 2026 State of AI in the Enterprise survey, covering 3,235 senior leaders across 24 countries, found 74% of organizations hope to grow revenue through AI initiatives, while only 20% currently do – nearly the same magnitude of gap as BILL's 82%-to-25% chain, measured a completely different way, on a much broader population. BCG's 2025 global AI value research found just 5% of companies worldwide are "future-built" – systematically capturing substantial AI value at scale – while 60% report reaping almost no material value despite real investment.

None of these three studies measured the same thing, or the same population. BILL surveyed accounting-firm leaders about client-facing expectations; Deloitte surveyed enterprise leaders about revenue outcomes; BCG measured value-capture at the organizational level, globally, across industries. I'm not treating them as one dataset – they're independent confirmations, from different methodologies, of the same underlying shape: expectation and ambition around AI are running well ahead of anything firms can currently point to and call a return. When three separately-run studies with three different questions all land on roughly the same gap, that's a pattern worth taking seriously even though none of them individually proves it applies to your specific segment.

What this rhymes with in boutique advisory

I'd apply this to strategic advisory and competitive-intelligence firms as an analogy, not as measured fact – nobody surveyed your segment specifically. But the rhyme is close enough to be useful. A client who's used AI tools themselves now has a felt sense of what "fast" and "current" should mean, and they bring that expectation into every engagement whether or not you've built anything to match it. That's the same pressure BILL is describing, just aimed at a different kind of deliverable. And the finding that even the best-converting AI initiative only pays off for a quarter of the firms that build it should be a caution against assuming that simply announcing an AI-powered offering closes that gap. Most of the time, per this data, it doesn't – you still have to actually deliver the substance behind the story.

That's the same dynamic I've pointed to in why most AI ROI gaps are a culture problem, not a tools problem: buying or building the capability was never the hard part. Getting an organization to actually change how it works around that capability is. It also connects directly to a risk I've written about from the other side of the table: when the AI story runs ahead of the AI substance, that's exactly the gap due diligence has had to get sharper about catching in acquisition targets. The same overclaim risk that shows up when a company is being evaluated for sale shows up, in a smaller and more everyday way, whenever a boutique firm tells a prospective client it's "AI-powered" without a specific, honest account of what that actually means for the work.

I don't think that's a coincidence, either. Both are versions of the same failure: a claim about AI capability that outruns what's actually been verified or built. One version gets caught in a deal room, with lawyers and auditors checking the gap. The other version gets caught, much more quietly, the first time a client asks a specific follow-up question about how the AI-powered part actually works – and the answer turns out to be thinner than the pitch.

What I'd actually do with this

If you run a small advisory practice, I don't think the answer is to rush out an AI-branded service line to meet the expectation BILL's survey describes – the data says that specific move converts to a real return only about a quarter of the time, even for firms with real resources behind it. The more useful move is being concretely honest with clients about where AI actually touches your process today and where it doesn't, rather than letting an ambient industry expectation set the bar you then have to perform against. The competitive pressure boutique advisors are actually facing isn't really about who has the flashiest AI feature – it's about whether clients still see a reason to pay for judgment once the story around AI capability has become table stakes everywhere. Matching the story to the substance, honestly, is a smaller ask than building a new service line, and per this data, it's also the one more likely to survive contact with an actual client relationship.