How the Index is made, and who makes it
Who publishes this
The Automation Index is written, funded and run by CXJ Labs, a company that builds automations for businesses. It is not independent and does not pretend to be. Every page says so. We link to our own services, and we would benefit if you hired us. Judge the estimates on their method, which is published below.
We do this openly because the US Federal Trade Commission's 2024 rule on reviews and testimonials bans company-run sites that claim to be independent reviewers, and because readers deserve to know. The Index does not review or rank vendors, takes no payment or commission from any tool it names, and has no affiliate links.
How we estimate the share that is automatable
We break each process into the four to seven tasks that make up the work, checked against O*NET task statements. Each task is mapped to one of McKinsey Global Institute's activity types, which gives a starting automation rate. We then adjust that rate for the specific task, based on what current tools can reliably do, and write down why.
| McKinsey activity type | What it means here | Base rate |
|---|---|---|
| Predictable physical work | Repeating the same physical action in a known setting. | 81% |
| Processing data | Moving, copying, checking, calculating or formatting information. | 69% |
| Collecting data | Gathering information from people, forms, inboxes or systems. | 64% |
| Interacting with people | Conversations where tone, trust or negotiation matters. | 20% |
| Applying expertise | Judgement calls, diagnosis, creative or strategic decisions. | 18% |
| Managing others | Coaching, prioritising and holding people to account. | 9% |
“Automatable today” means a small business could do it now with off-the-shelf tools: n8n, Make, Zapier, GoHighLevel, mainstream CRMs and accounting software, and AI models used inside those workflows. It does not mean every business should, and it does not count capabilities that are only demonstrated in research.
How we estimate hours saved
Each entry is modelled on one stated scenario, such as a team booking 60 sales calls a week. We estimate the minutes each task takes by hand in that scenario and list the assumptions. Hours saved is each task's minutes multiplied by its automatable share, summed, divided by 60 and rounded to the nearest half hour. Change the scenario and the hours change.
Across all 50 entries the median process is 69%automatable. That figure describes our scenarios, not the economy. It sits close to McKinsey's 2023 estimate that current technology could automate activities absorbing 60 to 70% of employees' time, which is a useful sanity check rather than a confirmation.
What the numbers are not
These are editorial estimates, not survey results or measured outcomes. They are rounded to the nearest 5% per task because more precision would be false. They ignore set-up time and running costs. They assume the process is already sensible, which is often untrue: automating a bad process makes it fail faster. Statistics from third parties are quoted with their caveats.
- We do not publish client results or case studies here, and no entry is based on a named customer.
- Vendor statistics are labelled as vendor statistics. Small or old studies are labelled as such on the sources page.
- Tool mentions describe a job a tool can do. They are not rankings or endorsements.
Updates and corrections
Each entry shows the date it was last reviewed. We revisit the estimates when tools change what is reliably possible, and at least every six months. If you think an estimate is wrong, tell us through cxjlabs.io with your reasoning. We will correct errors on the page and note the change, whether or not it flatters automation.
The data is free to reuse under CC BY 4.0. Download it as JSON or CSV, and see every reference on the sources page.