Before You Buy Another AI Tool
About this resource
The average organization is now paying for AI capability it already owns twice. Features arrive inside software you already license, a department buys a point solution nobody else knows about, and a renewal goes through on a seat count nobody has audited. This free checklist covers the ten questions to answer before an AI tool purchase and the eight to put to the vendor.
The internal questions cover what problem this solves, what it costs today, whether you already own something equivalent, who will own adoption, and what happens if usage never materializes. The vendor questions cover data handling, retention, training opt-out, exit and portability, and what happens when their terms change. It includes a table for defining success before the purchase rather than after it, and it closes with four honest options, one of which is doing nothing, which is a legitimate outcome and rarely written down as one.
Buy, pilot, use what you already own, or do nothing. The fourth is a legitimate decision and almost never gets recorded as one, which is why the same proposal reappears next quarter.
How to use it
Work the ten internal questions before you take the vendor call, because the answers change what you need from the conversation. The single highest-return question is whether you already own something equivalent; features keep arriving inside platforms organizations already license, and nobody is tracking them. If you cannot state what the process costs today, the request is premature regardless of how good the tool is.
Inside the download
The feature may already be inside a platform you license, which is the most common finding.
What the process costs today, because without it there is nothing to compare against.
Who is responsible for usage, by name, before money moves.
Where data lives, for how long, whether it trains their models, and what happens on termination.
Common questions
Internally: what specific process does this improve, what does that process cost today, do we already own something that does this, who owns adoption by name, what does success look like and by when, and what happens if usage never materializes. Externally: where does our data live, for how long, does it train your models, what happens on termination, and what portability do we have.
Where data is held and under whose jurisdiction, how long it is retained, whether it may be used to improve their models and how to opt out, whether the answer differs by subscription tier, and what happens to your data on termination. Get the answers in the contract rather than from a settings menu, since settings can be changed by somebody who does not know what they are changing.
Yes, and it should be written down as a decision rather than left as an absence of one. Doing nothing is legitimate when the process cost is unknown, when nobody will own adoption, or when an existing platform already provides the capability. Recording it prevents the same proposal returning next quarter as though it had never been considered.
Free. No email list, no follow-up sequence.
More from the resource library, chosen because they pair well with this one.
Why most AI return is undefendable, and the seven questions that have to be settled before any money moves.
Ten questions for directors approving something they cannot evaluate, with what a good answer sounds like and what should concern you.
The criteria that should decide an organizational AI assistant: licensing, data terms, account type, integration, and support. Written for the organizational decision rather than the individual one.
Tell us a little about yourself and this document is yours. We don't have an email list. We collect this so we know who's finding our resources useful, and that's it.
You're downloading
this document
Your information is kept confidential and never shared with anyone.
