Overlapping Software: When to Drop a Tool That Still Works

Most advice about switching software assumes the current tool is bad. That case is rare and easy: when something is failing, the decision makes itself. The hard case is a tool that works, that nobody complains about, and that another product you already pay for has quietly started to duplicate. This article is a method for that case. It uses vendor documentation for one common overlap, design review and client approval, and a blank worksheet you can apply to any pair of tools.

Correction, September 25, 2026: an earlier version of this article described a three-year subscription to a client proofing product, a four-month delay, two annual approval mix-ups and a six-week client notice period. None of that was a documented event, and the figures were not measured. This version replaces the story with published product documentation, cited research on why overlapping tools are kept, and a procedure you can run yourself. Mirror has not run the procedure on a customer account.

two overlapping circles of a Venn diagram slowly merging

Illustration only.

Why a working tool is the hard case

Two well-documented effects push in the same direction. The sunk cost effect is the tendency to continue an endeavour once money, effort or time has been invested in it. Arkes and Blumer’s 1985 paper documented it in experiments and in a field study of theatre subscribers: people who had paid more for a season ticket attended more plays over the following six months, apparently because of the higher sunk cost rather than any difference in the plays. Years of a tool working are a sunk cost, and they feel like evidence about the future. They are not.

Status quo bias is the separate tendency to prefer the current arrangement when a choice is framed as changing versus not changing. Samuelson and Zeckhauser’s 1988 experiments found the bias grows with the number of alternatives on the table. The subscription you meant to cancel covers that literature and its limits in more detail. For overlapping tools the relevant point is narrow: “keep both” is the status quo, so it wins ties by default, and the tie is never actually scored.

The scale of the overlap problem is documented at the enterprise level. Zylo’s 2026 SaaS Management Index, published January 29, 2026, reports that the average organisation in its dataset manages 305 SaaS applications, adds an average of 21 applications a month in large enterprises, and leaves 36% of licences unused. Generative AI appeared on its “most redundant app functions” list for the first time, at an average of seven apps per portfolio. Those are figures from large companies whose spend Zylo manages, not from small teams, so treat them as a description of the direction of drift rather than a benchmark for your own stack. The mechanism is the same at any size: products grow toward each other, and the second tool’s overlap with the first only ever increases.

A documented overlap: client review inside the design tool

Dedicated proofing and approval products exist because design tools historically had no way to collect client feedback. That is no longer true, which makes the pair a clean example. The table records what each vendor documents today. It is not a measured comparison of how well each one works.

Vendor documentation checked September 25, 2026. Prices are USD list prices from the vendor pricing pages on that date; regional pricing, taxes and promotions vary.

Product	What the documentation says	What it does not cover	Source
Figma comments	Anyone with at least “can view” access to a file can add and view comments, on any team or plan. Limit of 100 comments per hour per user; up to 5 images or GIFs per comment.	A multi-stage approval workflow, a formal sign-off record, or review of non-Figma files such as PDFs and video.	Figma Help: Comment in files
Filestage	Free plan: 1 active project, 5 new files a month, 2 reviewer groups, 2 GB. Starter $199 a month with 10 users included and unlimited reviewers; Business $329 a month.	Whether your clients will comment in a second place at all; the plan’s value depends entirely on review volume.	Filestage pricing
Ziflow	Personal plan free: 2 users, 2 GB, 1 workflow stage, 60 days of review history. Standard $199 a month billed annually, 15 users, 2 workflow stages; Pro $329 a month, 20 users, 3 stages. Unlimited reviewers on all plans.	Same as above; the free tier’s 60-day history limit matters if approvals are part of your audit trail.	Ziflow pricing

Read across the rows and the real question appears. The design tool’s built-in comments cover the conversation about a design. The dedicated tools sell the workflow around the conversation: stages, versions, a sign-off record, and review of file types the design tool does not open. If your work never leaves the design tool and “approved” is a client saying so in a comment thread, the overlap is nearly total. If you deliver PDFs and video, or a client’s legal team needs a dated approval record, the overlap is partial and the dedicated tool still does a job.

That distinction is what the worksheet below is for. It replaces “do we like it” with “which jobs does each tool do, and which record is authoritative for each.”

The replacement test

The most useful reframing I know is to stop asking whether the tool is good and ask instead: if we did not have this today, would we buy it at this price, given everything else we now own? This is a framing device, not a measured technique, and it has a known weakness: it is easy to answer “no” in the abstract and then discover a dependency in week two. The dependency test in the tools you cannot cut exists to close that gap before you cancel anything. It is also the same question the quarterly stack audit applies to every line item.

Price the duplication, not the tool

A subscription price is easy to justify on its own. The cost that never appears on an invoice is the cost of one job being split across two records. Put a number on it, and label the number as your estimate.

The arithmetic below is hypothetical, to show the shape of the calculation. Replace every figure with your own.

Dedicated proofing tool at the Starter tier: $199 a month, or $2,388 a year.
Cost of split records: suppose two approvals a year land in the wrong system, each takes four hours to reconcile, and the hours are worth $60 each. That is 2 × 4 × $60 = $480 a year.
Cost of client friction: a second login and a second place to comment for every client. Unpriced, but it is the item clients complain about.

In that made-up example the duplication costs $480 a year and the subscription costs $2,388, so the question is whether the dedicated tool’s remaining unique jobs (multi-stage approval, non-design files, a sign-off record) are worth $2,388 a year to you. If those jobs do not exist in your work, the tool is a decision you have already made and not yet executed. If they do exist, keep it and stop treating it as duplication. Either answer is fine. Not scoring it is the only wrong outcome. For a per-use view of the same logic, see cost per use for subscriptions.

Overlap worksheet

One row per job the two tools both claim. Fill it from your own account, not from marketing pages.

Job	Tool A does it?	Tool B does it?	Authoritative record today	What breaks if only one remains	Decision
Client comments on a design	Yes, in-file	Yes	Split	Nothing structural; one place to look	Consolidate in A
Formal sign-off with date	No	Yes	B	Audit trail of approvals	Keep B or replace with a written email approval
Review of PDF deliverables	No	Yes	B	PDF review moves to email	Decide per client
Review history older than 60 days	Yes	Plan-dependent	Check	Recovering an old approval	Export before cancelling

The rows are examples for the design review case. The columns are the reusable part. The “authoritative record” column is where overlap does its damage: when both tools hold half of the truth, the cost shows up as a reconciliation task months later, and nobody books it to either subscription.

Leave reversibly

If the worksheet says consolidate, do it in a way you can undo.

Export the record before cancelling, and open the export somewhere other than the tool that produced it. What the export actually contains has a procedure for checking coverage and attachments, and what vendors keep after you cancel covers the deletion clock.
Run both for one cycle with a rule about which one is authoritative. The trial is not “use both”, which is the current problem. It is “new work goes only to A; B is read-only.” If a job cannot be done in A for a real client, you have found the dependency the replacement test missed.
Move billing to monthly first. If the tool you are leaving is on an annual plan, the annual discount is a loan explains why the prepaid months are not a reason to stay and how to time the exit at the renewal date.
Tell the people who comment. Clients and colleagues need the new location and a date. The rollout advice in migrating eleven people off a tool applies even when the “team” is external reviewers.
Apply the check at purchase next time. Before adopting anything, list what it overlaps with and whether the overlap is likely to grow, alongside the pre-trial pricing audit. If the answer is yes, you are not buying a tool. You are scheduling a migration and choosing to pay for two things until then.

Checklist

Name the two tools and list every job both claim.
Mark which record is authoritative for each job today. “Both” means neither.
Read the vendor documentation for what the remaining tool actually supports on your plan, not the plan above it.
Price the duplication with your own hours and rate, and write the figures down as estimates.
Ask the replacement question, then run the dependency test before acting on the answer.
Export first, run one cycle with a single authoritative tool, and cancel at the billing boundary.

Sources

Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124–140. Publisher record
Samuelson, W., & Zeckhauser, R. (1988). Status quo bias in decision making. Journal of Risk and Uncertainty, 1, 7–59. Publisher record
Zylo, 2026 SaaS Management Index announcement, January 29, 2026, and the supporting statistics page. Checked September 25, 2026.
Figma Help Center: Comment in files. Checked September 25, 2026.
Filestage pricing and Ziflow pricing. Checked September 25, 2026.