.svg)
Most automation projects don’t fail because the technology was wrong. They fail because the wrong process got automated first.
Pick a low-frequency process, and the savings never compound. Pick a low-impact one, and nobody notices when it works. Pick one that needs human judgment in every case, and you spend three months building a system that breaks the moment a real customer uses it. Three failed automation projects later, leadership concludes “automation doesn’t work for us” when the real problem was the order of operations.
This guide shows you how to identify which business processes to automate first using a simple scoring framework called FIT (Frequency × Impact × Tractability). It takes about 90 minutes to apply across an entire business. By the end, you’ll have a ranked list of automation candidates, the top three pressure-tested with the people who actually run them, and exactly one process to start building this quarter.
Why “Automate Everything” Is the Wrong Strategy
The biggest mistake teams make when starting automation is treating it like a buffet. They list 30 processes, get excited, try to tackle five at once, and three months later have five half-finished projects and zero working systems. Automation compounds when it’s done in sequence, not in parallel, and only when each project actually ships.
The second mistake is automating what’s annoying instead of what’s expensive. A monthly board report that takes a full afternoon is annoying, but it runs 12 times a year, so the upside is capped. A lead intake process that runs 30 times a week and costs you a deal every time someone forgets to follow up is expensive, and the upside is enormous. The goal isn’t to remove what bothers you; it’s to remove what’s costing you.
The framework below sorts through both problems by scoring every process on three dimensions that, multiplied together, predict automation ROI better than any single criterion alone.
The FIT Framework: Frequency × Impact × Tractability
FIT scores any business process on three independent dimensions, then multiplies them. Multiplication matters: a process scoring 1 on any dimension is dragged to a low total even if the others are high. That’s the whole point. A process that runs once a year can’t be a top automation candidate, no matter how painful it is.
F - Frequency: How Often Does It Run?
Frequency is the multiplier that turns small per-event savings into large annual ones. A 5-minute task automated once a year saves 5 minutes. The same task, automated when it runs 50 times a day, saves over 200 hours a year.
Frequency tends to be undervalued by teams new to automation. The boring, repetitive, high-volume processes are almost always the highest-ROI targets; they’re just less interesting to talk about than the rare, high-stakes ones.
I - Impact: What Does Failure Cost?
Impact captures the cost of getting the process wrong, not the cost of doing it manually. A process can be tedious and high-frequency but low-impact; those are still worth automating, just not first. The processes worth tackling first are those where errors translate directly into lost revenue, customer churn, regulatory exposure, or time-sensitive opportunities.
The best test for impact is asking: “If this process failed silently for a week, what would it cost the business?” If the honest answer is “nothing visible,” the score is 1. If the honest answer involves a number with three or four zeros, the score is 3.
T - Tractability: How Hard Is It to Automate?
Tractability is the dimension most teams forget to score, and it’s the one that kills the most projects. A process can be high-frequency and high-impact but requires so much human judgment that any attempt to automate it produces worse outcomes than leaving it alone. Tractability asks: “How rule-based is this work?”
Tractability is also the dimension that AI has changed the most. Five years ago, scoring 2 on tractability often meant “don’t automate.” With LLMs in the loop, a lot of formerly fuzzy work is now firmly automatable: classifying tickets, summarizing emails, and extracting data from invoices.
But there’s still a hard ceiling. Anything requiring genuine judgment under uncertainty (negotiating contracts, designing products, hiring) still scores 1, and trying to automate it usually wastes money.
The Formula: F × I × T = FIT Score
Multiply the three scores together. The result lands between 1 and 27, and the action depends on the band.
Worked Example: Scoring 6 Common Processes
The framework is easier to internalize when you watch it work. Here’s FIT scoring applied to six processes a typical mid-market business might consider automating. Same business, same team, very different priorities.
A few things are worth noting from this exercise.
Lead intake wins decisively. It’s the most frequent (multiple leads per day); highest-impact (every minute of delay costs deals; research on lead response consistently shows that replying within minutes rather than tens of minutes dramatically lifts conversion); and most tractable (clear inputs, deterministic routing rules). Score: 27. This is the textbook “automate first” answer for almost any business with inbound demand.
Invoice processing scores nearly as high. It runs multiple times a day; errors directly hit cash flow, and modern OCR plus LLM extraction makes it mostly rule-based, with some exceptions requiring review (Tractability 2). The score of 18 puts it firmly in the green band. For finance-heavy businesses, this often rivals lead intake as the first thing to build.
The high-impact trap is real. Notice that custom enterprise proposals score Impact 3; getting them wrong absolutely costs deals. But Frequency is 1 (a few per quarter), and Tractability is 1 (every proposal requires significant judgment). Total FIT score: 3. This is the trap that catches teams who automate based on “what hurts most” rather than what scores highest. Custom proposals hurt, and they should still be the last thing you automate.
Quarterly board reports are tempting and wrong. They feel like obvious candidates because the work is visibly tedious. But the FIT score of 4 tells the truth: low frequency means almost no compounding savings, and the actual time cost is a few hours per quarter. Automating it might feel satisfying, but the same effort spent on a 27-score process delivers far more return.
The 5-Step Process to Run This Across Your Business
Knowing the formula is one thing. Applying it across an entire business in a single afternoon is the actual deliverable. Here’s the 5-step process most teams use, with realistic time estimates for each step.
Step 1: Inventory (15 minutes)
Open a spreadsheet. List every recurring process across the business: finance, sales, operations, HR, customer support, and marketing. Don’t filter yet; don’t pre-judge what’s automatable. The goal is just a complete list. Most teams end up with 25 to 60 processes by the time they’re done.
If you’re stuck, ask each department lead one question: “What’s something your team does the same way every week or every day?” Their answers will give you 80% of the inventory in 10 minutes.
Step 2: Score (30 minutes)
For each process on the list, assign three numbers: F (1, 2, or 3), I (1, 2, or 3), and T (1, 2, or 3). Don’t agonize. The whole point of a 1–3 scale is to force quick directional judgments rather than precise estimates. If you can’t decide between two scores, default to the lower one; overweighting your gut feel will throw off the ranking.
Have one person score the full list, then have a second person review. Disagreements about scores usually surface useful information about how the process actually works, which feeds directly into Step 4.
Step 3: Rank (15 minutes)
Multiply the three scores per row. Sort the spreadsheet by FIT score, descending. The top three rows are your candidate list. Anything 18 or above is a serious candidate; anything below 8 should be cut from consideration entirely.
For most businesses, the top three end up being some combination of lead intake, invoice processing, support ticket routing, client onboarding, or recurring billing. The exact mix depends on whether you sell to consumers, businesses, or enterprises, but the top three almost always cluster in the same operational areas.
Step 4: Validate (20 minutes)
Take the top three to the people who actually run those processes. This step is non-negotiable; nothing kills automation projects faster than building a workflow that doesn’t match how work really happens. Ask each owner three questions:
• “What does this process actually look like, step by step?”
• “Where does it usually go wrong?”
• “What edge cases would break a simple version of this?”
If the answers reveal that a process is much more judgment-heavy than the score suggests, drop the T score and recalculate. If the process turns out to be mostly already standardized, the project is even easier than expected. Either way, validation surfaces the truth before you start building.
Step 5: Pick One (10 minutes)
This is the step teams skip, and it’s the one that determines whether the whole exercise produces results. Pick the highest-scoring process from the top three. Build only that one. Ship it. Get it stable. Then come back and pick the next one.
The temptation to start three projects in parallel is enormous, especially with a list of high-scoring candidates in front of you. Resist it. Three projects competing for the same engineering attention almost always produce three half-finished automations. One project finished produces one working automation, plus a playbook the team can reuse on the next one.
Three Common Mistakes That Wreck the Prioritization
Mistake 1: Scoring Frequency on Total Volume Instead of Per-Process Volume
A team scoring “customer support” as Frequency 3 because they handle 200 tickets a day is making a category error. Customer support isn’t one process; it’s dozens. Ticket triage runs 200 times a day (Frequency 3).
Refund approval runs several times a day (Frequency 3 as well, since that’s still multiple times per day). Complaint escalation runs once or twice a week (Frequency 1). Score processes, not departments, and apply the band definitions literally.
Mistake 2: Confusing Pain With Impact
“This process is so painful” is not the same as “this process has high impact.” Pain measures how much the work bothers the people doing it; impact measures what failure costs the business.
The two often correlate, but not always. The most painful work is often manual data entry: frequent, tedious, and hated.
The highest-impact work is often something that runs in the background until it doesn’t, like contract renewals or compliance reviews. Automate the highest impact first, even if the most painful work feels more urgent to your team.
Mistake 3: Underestimating Tractability for AI-Powered Workflows
The opposite mistake is also common. Teams that haven’t worked with modern AI tools score tractability conservatively because they remember 2020-era automation, when anything fuzzy was off-limits. With current LLMs, classification, summarization, extraction, and routing are all firmly tractable.
If you’re scoring something a 1 because it requires “reading and understanding text,” upgrade to a 2. AI has moved that line. The work that genuinely scores 1 today is work requiring strategic judgment under uncertainty, not work requiring language understanding.
Frequently Asked Questions
1: What business processes should you automate first in 2026?
The five processes most businesses should automate first, in rough order of typical FIT score: (1) lead capture and CRM sync, (2) invoice processing and approval routing, (3) customer support ticket triage, (4) client or employee onboarding workflows, and (5) recurring KPI reporting.
The exact order depends on your business model: B2B service businesses usually start with lead intake, e-commerce businesses with order processing, and finance-heavy businesses with invoicing, but these five cover roughly 80% of high-ROI automation work for most companies.
2: What is the FIT framework for automation prioritization?
FIT is a scoring framework that ranks automation candidates on Frequency × Impact × Tractability. Each dimension is scored 1 to 3. Frequency measures how often the process runs (1 = less than weekly, 3 = multiple times daily). Impact measures what failure costs (1 = mild inconvenience, 3 = lost revenue or churn).
Tractability measures how rule-based the work is (1 = heavy judgment, 3 = pure rules). The three scores multiply together, producing a FIT score from 1 to 27. Scores of 18 and above are automation priorities; scores below 8 should be skipped or have their underlying process fixed first.
3: Should I automate high-impact processes or easy processes first?
Neither in isolation. Automate the processes that score high on impact, tractability, and frequency at the same time, plus frequency. High-impact processes (like custom proposals or strategic decisions) often score low on tractability because they require judgment, which means automating them produces worse outcomes than leaving them manual.
Easy processes (high tractability) often score low on impact, which means the savings don’t matter much. The sweet spot is the intersection of high impact, high frequency, and high tractability, which the FIT framework captures by multiplying all three together.
4: How long does it take to identify automation candidates using this framework?
About 90 minutes for most businesses, broken down across five steps: 15 minutes to inventory all recurring processes across departments, 30 minutes to score each process on F, I, and T, 15 minutes to rank by FIT score, 20 minutes to validate the top three with the people who actually run them, and 10 minutes to pick the one to build first.
Larger enterprises with more processes may take a half-day; smaller businesses with under 20 candidate processes can finish in under an hour.
Where to Go From Here
Identifying what to automate is the first half of the problem. Building the automation well is the second half. Once you’ve run the FIT framework and chosen your highest-scoring process, the next questions are practical: which automation tool fits, how to design the workflow, what to monitor, and what to do when it breaks.
Whatever process you picked lead intake, invoice processing, support triage, onboarding, or reporting while there’s a playbook for building it well. The mistakes are fairly predictable, the patterns are repeatable, and the difference between a 60-day payback and a 6-month payback usually comes down to execution choices, not which tool you used.
Start with the highest score. Build it well. Get it stable. Then come back to your ranked list and pick the next one. Two or three FIT cycles a year, executed in sequence, will move a business from “automation feels overwhelming” to “automation is how we operate” within 12 to 18 months.
When You’d Rather Not Build It Alone
If your top-scoring process lands in the green band but your team is already stretched, this is exactly the kind of work Amplence takes on.
The FIT framework points at the high-frequency, high-impact, mostly rule-based processes that AI handles well today, and that’s the sweet spot Amplence builds for: customer service pipelines, document automation, data enrichment, and reporting workflows tuned to a single client’s data and decision logic rather than generic templates.
A useful reference point is CollageDepot, an e-commerce business whose support volume scored exactly the way support triage scores on FIT: high frequency, high impact, and tractable enough for modern LLMs.
Amplence built an AI-powered customer service pipeline that now handles 5,000+ emails a month across four languages, auto-resolving roughly 65% of them with response times under 60 seconds. That is the compounding-savings case the framework predicts: a process that runs constantly, where every hour of delay costs goodwill, automated cleanly because the inputs are structured and the rules are learnable.
The point of sharing it here isn’t the headline numbers; it’s that the project started the same way the framework recommends. One process, scored honestly and validated against how the work really happened, was then built well before anyone moved on to the next one. If you’ve run the FIT exercise and want a second set of hands on the highest-scoring candidate, that’s the conversation to have.
