The AI Readiness Checklist: 6 Things to Fix Before You Automate Anything

Open any foundation for the detail and the questions to ask yourself.

1. 01Data IntegrityIf nobody owns the ongoing quality of your CRM data, AI will scale every error already in it.+

   Most companies think their data problem is a tool problem. It's not. It's a governance problem.

   You can have the best CRM on the market and still not trust your pipeline data. The issue usually isn't the system. It's what's inside it. Duplicates. Missing fields. Conflicting data across objects. Stale records that nobody knows how to interpret because the person who set them up left two years ago. Fields that require manual entry but nobody fills out consistently because there's no automation doing it for them.

   But here's what gets overlooked most often: the real data integrity issue isn't any individual data problem. It's that there's no system or process in place to continuously vet, correct, and flag the data that matters most. No governance cadence for reviewing account, opportunity, or quote data. No automated alerts when critical fields are missing or conflicting. The data degrades by default, and nobody notices until the board deck doesn't add up.

   What good looks like vs. what we usually find:

   	What good looks like	What we usually find
   Pipeline trust	Leadership pulls the pipeline report and acts on it as-is	Every report gets manually checked (and "fixed") before it goes to the CEO
   Duplicates	A defined rule decides which record wins, and merges happen on a schedule	Duplicates pile up until a rep complains or a sequence emails the same person twice
   Required fields	Critical fields are populated automatically or enforced at the stage gate	Fields depend on reps remembering, so completion is inconsistent
   Governance cadence	Someone owns data quality and reviews it weekly or monthly	Nobody owns it; cleanup happens in a panic before board meetings
   Alerts	Missing or conflicting critical data triggers an alert before it compounds	Errors are found by hand, after the damage is done
   Definitions	Every critical field has a written definition, owner, and source	Fields mean different things to different teams

   The data integrity questions that matter:

   • If your CEO asked for a pipeline report right now, would you trust the numbers without manually checking them first?
   • Do you have a defined process for identifying and resolving duplicate records, missing fields, and conflicting data across your core objects (accounts, contacts, opportunities, quotes)?
   • Is there any automated system that flags data quality issues before they compound, or does your team find errors manually, after the damage is done?
   • Could you explain to a new hire how every critical field in your CRM gets populated, by whom, and how often it's validated?

   If you answered "no" to two or more of these, your data foundation isn't ready for AI. Any automation you build on top of it will inherit every problem your data already has, and scale those problems faster than your team can fix them.

2. 02Process DocumentationYou can't automate a process nobody has written down, and AI turns good documentation into a working asset.+

   Process documentation used to feel like busywork. Write the SOP, put it in a folder, watch nobody read it. AI changes that. Documented processes can be loaded into knowledge agents, custom GPTs, and AI assistants, and suddenly your team can ask questions and get precise, contextual answers directly from the documentation instead of digging through 40-page manuals nobody wants to open.

   That reframe matters because it turns process documentation from a compliance exercise into an AI-readiness investment.

   The symptoms of undocumented processes are predictable. Critical processes live in one person's head. Two reps do the same thing completely differently. Handoffs between teams are inconsistent because nobody agreed on what triggers them or what information needs to transfer. New hires take months to ramp because tribal knowledge is the only knowledge. And, connecting straight back to data integrity, bad data gets produced in your systems as a direct result of processes being undocumented and executed inconsistently.

   How documented are your processes, really?

   • If your top-performing rep left tomorrow, could someone else follow their exact process from lead to close without calling them?
   • Are your critical workflows (lead routing, opportunity management, deal handoffs, renewal processes) documented in a way that a new hire could follow them on day one?
   • Do different team members execute the same process the same way, or does the output depend on who's doing it?
   • Have you updated your process documentation in the last six months, or is it a snapshot of how things worked two years ago?

   If your processes aren't documented, you can't automate them. And if they're documented but outdated or inconsistently followed, automating them just codifies the inconsistency. AI can't improve a process that nobody has written down.

3. 03Systems ArchitectureA tech stack that was accumulated instead of designed will fight every automation you try to build on it.+

   There's a difference between a tech stack that was designed and one that was accumulated. Most growth-stage companies have the latter: tools bought by different teams at different times to solve different problems, with no one ever stepping back to ask whether they work together as a system.

   The most obvious signal is integration quality. Sometimes the systems aren't integrated at all. Data lives in silos and people manually copy it between tools. Other times there are integrations, but they're poorly mapped or incomplete. Fields don't sync correctly. Records duplicate across systems. Someone set up the integration 18 months ago and nobody has touched it since.

   The deeper issue is whether you have a single source of truth. For any given data object (a customer, a deal, a subscription), can you point to one system that owns it? Or do you have three tools that all think they're the source of truth, with no clear hierarchy?

   This is one of the most damaging problems a growth-stage company can have. We've seen a company run its entire go-to-market on a CRM that wasn't built for the scale or complexity the business had reached. It barely integrated with anything: not the financial system, not the marketing platform. Even modules inside the CRM didn't properly talk to each other. The result was operational chaos: enormous manual effort, strange workaround processes, data nobody believed, and go-to-market results that dragged as a direct consequence. The frontline employees doing all that manual entry were miserable and disengaged, because their jobs felt pointless. That's what an accumulated tech stack does to a business. It doesn't just create data problems. It creates people problems.

   Beyond integration, watch for these signals: redundant tools doing the same job because different teams bought their own solutions. Manual data entry bridging gaps between systems that should be connected. No clear owner of the stack as a whole, where individual tools have admins but nobody is responsible for how the entire system works together. And tools used at a fraction of their capability: paying for enterprise-tier software but using it like a spreadsheet because nobody configured it properly.

   Pressure-test your systems architecture:

   • Can you draw a diagram of how data flows between your core systems (CRM, marketing automation, CS platform, billing), and does that diagram reflect what actually happens, or what you think happens?
   • For your most important data objects (accounts, opportunities, customers), is there one clearly defined source of truth, and does every team know what it is?
   • How many of your integrations were set up more than a year ago and haven't been audited since?
   • Are you paying for tools that overlap in functionality because different teams bought their own solutions?

   A tech stack that wasn't designed as a system will fight you every step of the way when you try to automate. AI needs clean data flowing through connected systems. If your architecture has gaps, your automations will too.

4. 04Team AlignmentIf your teams don't share definitions and handoff rules, AI surfaces the disagreement faster and at higher volume.+

   This one is deceptively important. You can have clean data, documented processes, and well-integrated tools, and still fail at AI implementation because your teams aren't aligned on the fundamentals.

   The most common alignment failures happen at the seams between teams. Marketing and sales can't agree on what counts as a qualified lead. Pipeline stages have names that don't clearly describe what's happening in them, and the criteria for moving a deal from one stage to the next aren't defined. Sales sells things that finance and accounting don't want to support, and RevOps gets stuck in the middle trying to make the deal structure work for everyone.

   We've lived this directly. Finance and sales at each other's throats, with RevOps caught in the middle, trying to support six-figure deals structured in ways that made them unprofitable. Nobody was exactly wrong. Sales was trying to close business and finance was trying to protect margins. But the misalignment between them created deals that hurt the company even when they "won."

   Pipeline stages are another constant source of pain. We've seen stages with names like "Engage" and "Catalyze" that nobody could define. Reps bucketed deals inconsistently, which made forecasting impossible. Worse, when bookings underperformed, sales leaders told their teams to create pipeline just to create it, not because the deals were real, but because pipeline volume felt like air cover for poor bookings. With no codified entry and exit criteria, nothing stopped junk from entering the pipeline. That's what happens when pipeline definitions are vague and advancement criteria aren't enforced.

   Handoffs are another friction point. CS asks customers the same questions sales already asked. Information captured during the sales process doesn't transfer cleanly to the team that needs it next. The customer experience feels disjointed because internally, it is.

   And then there are the definition problems. What does "closed-won" actually mean: when the contract is signed, when the first payment arrives, or when onboarding starts? What counts as an MQL versus an SQL? These sound like semantic debates, but they create real measurement problems. If your teams measure different things and call them by the same name, your data will reflect that confusion, and any AI you build on top of it will amplify it.

   Where does team alignment actually break down?

   • Do your marketing, sales, and CS teams share a common set of definitions for lead stages, pipeline stages, and customer lifecycle stages, and are those definitions written down?
   • When a deal moves from sales to onboarding, does the receiving team have everything they need without asking the customer to repeat information?
   • Is there clear agreement between sales and finance on deal structures, discounting authority, and what can and can't be sold?
   • Could you explain your pipeline advancement criteria to someone outside the company in under two minutes, and would every member of your revenue team describe it the same way?
   • If pipeline creation is underperforming, can your sales leaders create deals that probably shouldn't be there, or do your entry criteria prevent that?

   Misalignment between teams creates noisy data, inconsistent processes, and conflicting metrics. AI doesn't fix alignment problems. It surfaces them faster and at higher volume.

5. 05Measurement FrameworkIf you can't work backward from your bookings target to weekly pacing, AI will just measure the wrong things more efficiently.+

   This is the one that separates companies that are truly ready to scale from companies that are tracking dashboards and hoping for the best.

   The right approach starts with the end in mind. What do you need to produce in bookings this quarter? Work backward from there, mathematically, using your actual conversion data. How many deals need to be in the pipeline, and at what value? How many qualified meetings does that require? How many MQLs feed those meetings? Each stage has a conversion rate, and if you don't know those rates from real data, that's your first problem.

   Once you have those numbers, the discipline is tracking your pacing against them continuously. If you're halfway through the quarter, you should know exactly whether you're at 50% of each metric, where you're short, and what the specific bottleneck is. Not at the end of the quarter. Every week.

   What we see more often is the opposite. Teams track metrics but don't act on them. Dashboards exist but nobody changes their behavior based on what they say. People report on activity volume (calls made, emails sent) without connecting it to the outcomes that matter. Many companies track too many metrics at once, diluting focus, instead of identifying the three to five numbers that drive the outcomes they care about and watching those relentlessly.

   There's also a measurement shift many companies haven't made yet. The growth-at-all-costs days of 2021 are over. Efficiency is at a premium. Companies should be measuring not just growth but efficiency metrics that show how they're progressing with AI and operational scalability. Think ARR per full-time employee. Cost per MQL. Cost per opportunity. LTV to CAC. These tell you whether your growth is sustainable and whether your operational investments are creating leverage, or just creating more cost. If you're not measuring how efficiently you grow, you're only seeing half the picture.

   Are you measuring the right things?

   • Can you work backward from your bookings target to the exact number of MQLs, qualified meetings, pipeline deals, and pipeline value you need at each stage, and are those numbers based on real conversion data?
   • Do you track your pacing against those targets weekly, and does your team take action based on what the data shows?
   • If you asked five people on your revenue team what the three most important metrics are, would they give you the same answer?
   • Are you measuring efficiency metrics like ARR per FTE, cost per MQL, cost per opportunity, and LTV to CAC, or just top-line growth numbers?
   • Are you measuring the things that drive outcomes, or the things that are easiest to measure?

   If your measurement framework isn't connected to the outcomes you're trying to drive, AI will just help you measure the wrong things more efficiently. The foundation here isn't more dashboards. It's knowing exactly what to measure, measuring it consistently, and acting on what it tells you.

6. 06AI Adoption ReadinessClean data and documented processes still fail if nobody owns AI and the team never changes how it works.+

   This is the foundation most checklists leave out, and it's where a lot of AI investment quietly disappears. A company can get the first five right and still stall, because the tools arrive and the habits don't change.

   The pattern is familiar. Individuals experiment on their own, a few people get real value, and nothing spreads. Leadership is interested but nobody owns the agenda, so there's no budget, no priorities, and no way to tell whether any of it is working. When a new tool rolls out, adoption spikes for a few weeks and then fades as everyone drifts back to the old way.

   Adoption is a project, not a byproduct. It needs an executive sponsor with a short list of measurable priorities, simple guidelines for which tools to use and how, training tied to real work, and someone in each team who champions the new way of doing things. Without that, even well-built automations sit unused.

   Is AI changing how your team works?

   • Is there an executive who owns AI, with a budget and two or three priorities you could name?
   • Does your team use AI as part of how work gets done, with approved tools and shared guidelines, or are people experimenting on their own?
   • When you last rolled out a new tool or process, was adoption still there six months later?
   • Do you measure adoption, or do you assume it?
   • If your most capable AI user left tomorrow, would what they built and learned stay with the team?

   If AI use depends on a few enthusiasts and nobody owns the outcome, the foundation isn't ready, no matter how clean the data is. AI only pays off when the people using it change how they work.