.jpg)
A $20M company sits across the table from three AI agencies they're evaluating. They have profit centers in tools like NetSuite and Hubspot, no data infrastructure, and are mid-migration from Microsoft to Google. They're asking about sequencing, data readiness, and organizational change. This is what AI adoption challenges actually look like in 2026, and 79% of organizations are facing them right now.
The objections companies raise about AI sound like they're about the technology. They're about data infrastructure, organizational readiness, and the order of operations. When a company says "we're not sure you can do this," what they mean is "we don't have the foundation to support it." When they say "our data is a mess," they're describing a data preparation problem that blocks every AI initiative. When they say "we need help with the order of operations," they're asking for a partner who can sequence the work.
We had this conversation with a $20M company. Every objection pointed to the same root cause: the barriers to enterprise AI adoption are organizational, and the technology is ready before the company is.
Every company evaluating AI consulting services faces the same problem: there are no benchmarks. No standardized way to compare one agency's capability against another. The prospect told us they were looking at three agencies and had no framework for measuring whether any of them could deliver.
This is one of the most common AI adoption challenges. Companies know they need AI. They don't know how to evaluate who can build it. The result is a procurement process that defaults to comparing proposals instead of comparing outcomes.
What works instead is milestone-based delivery. A partner worth hiring should be able to define specific, measurable outcomes at each phase of the engagement. Concrete deliverables with success criteria, not vague promises about capability.
At NineTwoThree, we've delivered 24 of 27 projects with positive ROI, and our project success rate is 97%. Those numbers exist because we build to milestones. If an agency can't tell you exactly what they'll deliver and how they'll measure it, that's your answer.
For companies wondering how to evaluate AI agencies without joining the 95% failure rate, the framework is simple: ask for proof points, ask for milestone definitions, and ask for references from companies in your industry.
The prospect described their data situation plainly: 38 profit centers running through NetSuite, no centralized data infrastructure, and a migration from Microsoft to Google still in progress. AI can't work on that foundation because there's nothing stable for it to build on.
Gartner predicts that through 2026, organizations will abandon 60% of AI projects unsupported by AI-ready data. The data underneath AI is the actual blocker, and most companies haven't addressed it. A Riffon analysis of enterprise AI adoption found that the primary barrier is messy data infrastructure, and model capability is secondary. An AI agent is like a new employee with no tribal knowledge. If it can't find the authoritative source of truth across siloed systems, it will be ineffective.
We saw this firsthand with a healthcare organization managing 50+ care centers. Each center had its own P&L, its own SharePoint, and its own way of structuring data. Before any AI could be built, we had to map every data source, standardize formats, and create a single source of truth. The data mapping took months. The AI build came after.
This is why data preparation for AI matters more than model selection. Companies that skip this step end up with AI that produces confident wrong answers across every decision it touches. Gartner estimates that poor data quality costs organizations an average of $12.9 million per year, and AI scales those costs instead of fixing them.
For a deeper look at why data foundations matter, our post on building better data before building AI covers the practical steps.
The prospect was reorganizing profit centers, collapsing silos, and restructuring how the company operated. They described it as doing enterprise transformation and technology transformation simultaneously.
54% of C-suite executives say adopting AI is tearing their company apart, according to WRITER's 2026 Enterprise AI Adoption Survey. That's a double-digit increase from 2025. When you add organizational restructuring on top of AI adoption, the pressure compounds.
Most companies try to do both at once and fail at both. The ones that succeed sequence the work: stabilize the organization first, then layer in AI. Or they find a partner who can help them do both in coordinated phases rather than parallel sprints.
Our CEO, Andrew Amann, put it directly: "The ones that aren't hounding us for P&L results in the first three months are the most successful. They're playing a 2 to 3 year game, and they know the compounding value comes from getting the foundation right."
This is why creating an enterprise AI strategy before building anything matters. A strategy that accounts for organizational change, data readiness, and sequencing is worth more than any model.
The prospect's leadership team was budgeting for 12 months of AI investment with no expectation of short-term ROI. They were chasing compounding value over 2 to 3 years. This is unusual. Most companies want to see ROI in the first quarter, and most AI pilots end up in the valley of despair because of it.
The companies that succeed with AI think differently. They invest in foundations, data, and sequencing first. The ROI comes later, and it compounds. Consumer Reports saw a $20M revenue lift from their AI initiative. K&L Wines saved 14+ hours per day at 99% SKU accuracy. SimpliSafe was cited by their CEO at acquisition. Those results came from sustained, sequenced investment, not a 90-day pilot.
Only 29% of companies are seeing significant returns from generative AI, despite 59% investing over $1 million annually. The gap between investment and return is almost always a sequencing problem.
For a framework on measuring AI ROI that accounts for the long game, our guide on how to calculate the ROI of AI breaks down the metrics that matter.
This was the prospect's most honest question. They knew they needed data work, organizational change, and AI. They didn't know what order to do them in.
Our team walked them through the sequence:
| Phase | What happens |
|---|---|
| 1. Discovery and alignment | Map every data source, understand the business processes AI will touch, and define success metrics with stakeholders. |
| 2. Data preparation | Clean, standardize, and centralize data before building anything. This is where most projects fail. |
| 3. MVP and validation | Build a small, focused use case that proves the concept. Measure results against the success metrics defined in phase 1. |
| 4. Scale and iterate | Once the MVP delivers, expand to adjacent use cases. The compounding value starts here. |
Why 99% of AI projects fail is almost always a sequencing failure. Companies build AI before preparing data. They scale before validating. They measure ROI before the foundation is set.
The prospect described their aspirational vision: AI agents handling routine work across all 38 profit centers. That vision is achievable, but only if the order of operations is right. AI implementation myths often start with skipping steps, and that's where projects fall apart.
If you're unsure where your organization stands, find your AI readiness score to identify what to fix first.
The prospect was buying a partner who could guide them through sequencing, data readiness, and change management. The AI is the output. The input is organized, sequenced work.
This is what AI consulting services actually deliver. A partner who can look at a $20M company with 38 profit centers and no data infrastructure and say: here's the order of operations, here's what to fix first, here's how long it will take, and here's what the compounding value looks like.
WRITER's survey found that 48% of executives feel AI adoption at their company has been a massive disappointment. The disappointment comes from companies buying tools when they needed a partner. From building models when they needed to fix data. From chasing ROI when they needed to sequence the work.
Companies that partner with an AI development company instead of just hiring a vendor see different results. The partner handles sequencing, data readiness, and change management. The AI becomes a byproduct of getting the foundation right.
The conversation with this $20M company confirmed what we see across every engagement: the objections to AI are about data, organization, and sequencing. When companies say they're not ready for AI, they're usually right, and the fix is better foundations.
If you're evaluating AI agencies and hearing the same objections, the question is whether you have a partner who can help you fix the foundation first. That's what we do at NineTwoThree. We've launched over 160 projects by treating AI implementation as an engineering discipline, not a trend chase.
Contact us if you want a partner who can help you sequence the work, prepare the data, and build AI that delivers compounding value.
For more on building an AI-first organization, read about becoming an AI native company and adopting an AI-first strategy.
.jpg)
