AI Tool Overlap: How to Detect and Quantify It

AI tool overlap is the situation in which several paid tools cover the same use case for the same people. The company buys the same function two or more times: for example, a general-purpose assistant license for the whole marketing department and a second one, from another vendor, for the same users writing the same texts. You detect it by starting from use cases, not tools, and quantify it in euros by counting redundant seats.

What does overlap between AI tools really mean?

The definition has two conditions, and both are needed. The first is the same use case: drafting text, document summarization, coding assistance, data analysis, information search. The second is the same users: the people who appear in more than one tool for that task.

Two tools used by different departments for different tasks do not overlap. Two tools covering the same task in different teams indicate fragmentation, which is resolved by choosing a standard, not by revoking seats.

It helps to distinguish three cases:

• Full overlap: same person, same use case, two licenses. The less-used seat is a candidate for revocation.
• Partial overlap: the same person uses both tools for different tasks, but one of the tasks could be done with the main tool. Assess case by case.
• Included overlap: a tool already included in a paid contract is accompanied by a separate purchase for the same function. This is the easiest to eliminate.

Overlap differs from the unused seat, which is a license nobody uses. For that topic, see how to find unused AI seats. Here the seat is used, but redundant.

Why does overlap arise and accumulate?

The causes are organizational, not technical. Department budgets allow autonomous purchases, pilots become permanent without review, and each team picks the tool it knows. Nobody owns the overall picture, so nobody sees the duplication. In addition, vendors extend the features of their products, and a secondary tool becomes redundant without anyone noticing.

How to detect overlap by department and use case

The method starts from a matrix with use cases in the rows and departments in the columns. In each cell, note the tools in use and the number of people using them. Where a cell contains more than one tool, there is a candidate overlap.

The operational steps:

1. Collect the license list. For each tool you need the contract, annual cost per seat, number of assigned seats and renewal date. Include purchases outside central procurement, often reimbursed as department expenses.
2. Link licenses to departments. Aggregate by department and role, never by individual: the analysis serves to decide on the portfolio, not to monitor anyone.
3. Extract usage data. Vendor admin consoles expose activity data to varying degrees. Check what is available for each tool and align the observation window, for example the last 90 days.
4. Map use cases. A use case register is the best source. Failing that, a short questionnaire to department heads is enough for a first map.
5. Cross-reference people, tools and use cases. For each use case, count how many people have more than one active tool and how many really use both.
6. Classify each cell. Full, partial, included or justified overlap. Note the reason when someone can explain it.

The result is a list ordered by value: first the cases with many costly redundant seats. Keep the matrix as a living document, so each renewal starts from an up-to-date map and not from a blank page.

How to quantify overlap in euros

The basic formula is simple:

Overlap cost = redundant seats × annual seat cost of the tool that can be removed

First-year net savings are the avoided cost minus one-off transition costs (training, migration of workflows, management time).

The example below is hypothetical: the values are invented for illustration and do not represent real prices from any vendor.

A company has tool A as its standard and tool B purchased in parallel. B has 150 seats at a hypothetical annual cost of €360 per seat, so €54,000 of annual spend (150 × 360). For the use case "drafting and summarizing text" the picture is:

• Marketing: 120 people with A, of whom 80 also have B.
• Legal: 40 people with A, of whom 25 also have B.

The overlapping B seats are 80 + 25 = 105, which is 70% of B's seats (105 out of 150). Their annual cost is 105 × €360 = €37,800.

The comparison with department heads shows that 20 people use B for a task A does not cover, and must be kept. Redundant seats become 105 − 20 = 85, for an avoided cost of 85 × €360 = €30,600.

Hypothetical one-off transition costs: €40 per seat for training and workflow adaptation, so 85 × €40 = €3,400.

First-year net savings: €30,600 − €3,400 = €27,200. From the second year, savings are €30,600, because transition costs do not recur.

Two methodological notes. The calculation must be repeated for each use case, without counting the same seat twice. And the saving is real only if you actually act on the contract: a seat reduction mid-term may not be refundable, so align decisions with renewal dates.

What to do after measuring overlap

The measurement is there to support decisions. For each group of redundant seats there are four possible outcomes:

• Revoke the secondary tool's seat, with notice and department head verification.
• Reassign the seat to someone with an uncovered use case.
• Keep with a written justification and a review date.
• Consolidate the tools, when overlap is widespread and choosing a standard is cheaper than managing seat by seat.

Order of work: first overlaps included in contracts already paid, then full ones, finally partial ones.

Before revoking, check that users do not migrate to personal accounts: losing a company license can push people toward ungoverned tools. The topic is covered in shadow AI: risks and how to manage it.

After three months, measure again: adoption of the remaining tools, reactivation requests, team feedback. If reactivation requests are many, the overlap was partly justified and the rule needs correcting. Read the savings together with business results, following the method for measuring AI ROI.

Common mistakes in measuring overlap

• Starting from tools, not use cases. Two similar products on the market do not overlap if they serve different tasks in the company.
• Forgetting exit costs. Training, migration and contractual constraints reduce first-year savings.
• Analyzing people instead of roles. Data should be aggregated by department or role, both for methodological soundness and because of the rules on remote monitoring of workers.
• Not setting a periodic review. Without a quarterly check, overlaps rebuild themselves.

Where Delphis fits

Delphis is a vendor-independent platform that connects spend, adoption, impact and regulatory obligations. It reconciles costs with actual usage, identifies overlaps between tools by department and use case and simulates the savings of each consolidation scenario before you touch contracts. Data on people stays aggregated by role or department. To see how the analysis is set up, see the Spend page.