Employee suggestion program: how to run a suggestion system that beats the suggestion box

An employee suggestion program is a structured process for collecting improvement ideas from staff, deciding on them, and acting on the ones that are approved. It has an owner, a published response window, a review method, and a visible record of what got implemented. Anything without those four is a suggestion box with better marketing.

The distinction matters because the two things fail differently. A box fails quietly: ideas go in, nothing comes out, and within a quarter people stop bothering. A program fails loudly and expensively, because it made a promise about responding and then broke it in front of the whole company. That is worse than never having asked, and it is the single most common outcome. If you are not resourced to answer every submission, run something smaller and honest instead.

Most programs sit inside operations, HR or a continuous-improvement function, and the good ones are narrow. "How do we reduce changeover time on line three" produces usable suggestions. "Any ideas welcome" produces a list of things people dislike about the building. Scope is the cheapest lever you have, and it is set before launch. Where the ideas go afterward, and how they get compared against each other, is covered on our idea management software page.

A suggestion box collects. A suggestion program collects, decides, replies and implements. The box is a piece of furniture or a form; the program is an operating commitment with a named owner and a stated turnaround. Everything that makes suggestion schemes work lives on the decide-and-reply side, which is exactly the side a box does not have.

This is why swapping a physical box for an electronic suggestion box or a virtual suggestion box for employees changes very little on its own. Digital submission removes a small amount of friction, which is real but minor, since the reason people stop submitting is almost never that writing the idea was hard. It is that the last one vanished. A form that emails a shared inbox nobody owns reproduces the original failure at higher speed.

The useful test before you buy anything: name the person who will reply to submission number forty, and name the date by which they will have replied. If you cannot answer both, tooling is not your constraint yet.

Reply fast, reply to everyone, and give reasons for the noes. Participation is a response to evidence, and the evidence employees use is what happened to the last idea somebody submitted. A program that answers every suggestion within two weeks will out-participate one that offers a cash award and answers when it gets around to it.

The specific moves that work, in rough order of effect:

• Publish the response window and hit it. Two weeks is a reasonable promise for a first decision. A promise you keep is worth more than a shorter one you miss.
• Explain every no. A one-paragraph reason costs a reviewer three minutes and preserves the relationship. An unexplained rejection ends it, and that person tells their team.
• Ask narrow questions on a schedule. A standing open box goes quiet. A campaign with a defined problem and a closing date does not, because there is a reason to act this week.
• Show the implementations by name. One shipped idea with the suggester credited recruits more submissions than any poster campaign.
• Let managers nominate, not filter. If suggestions route through a line manager who can quietly kill them, you have built a program that measures how approachable managers are.

Recognition helps, and it is second-order. Cash awards in particular are less effective than the literature implies, because they turn a contribution into a transaction and invite arguments about how much an idea was worth. Attribution and a fast, reasoned answer do more work than a check.

Toyota runs the most cited employee suggestion system in the world, and its published numbers are worth quoting precisely because they are so often quoted loosely. Per Toyota own account of its Creative Idea Suggestion System, employees submitted around 810,000 suggestions in 2023, roughly 67,000 a month, averaging 14.4 per person across the year, and the system has gathered about 50 million suggestions across a history of more than seventy years.

Two things in that data get misread. The first is the size of the ideas. Toyota is explicit that these are mostly small improvements rather than breakthroughs, which is the entire point: a suggestion system is a machine for accumulating thousands of tiny operational gains, not for producing a new product line. Companies that launch a program expecting strategy back are measuring it against a job it was never built to do.

The second is that the paper-and-digital split, still roughly even in a company of that scale, undercuts the common assumption that participation is a software problem. Toyota gets 14.4 ideas per person per year with half of them on paper. What it has that most programs do not is seventy years of employees watching their suggestions get implemented. That is the asset, and no tool ships with it.

Offer anonymity as an option, do not make it the default. Named submissions are worth more to everyone: the reviewer can ask a clarifying question, the suggester gets credited when it ships, and the program builds the visible track record that drives the next round of participation. None of that is possible with an anonymous queue.

The option still has to exist, because some genuinely valuable suggestions carry risk for the person raising them. Anything that implies a manager is doing something wrong, anything touching safety, and anything that criticizes a decision made by someone senior will not get submitted under a name in most organizations. If you close the anonymous route you do not stop those observations existing, you just stop hearing them.

The workable arrangement is a named default with a clearly marked anonymous path, and an explicit commitment that anonymous submissions get reviewed on the same clock as the rest. That last part is where programs quietly cheat. Anonymous ideas are harder to process, so they drift to the bottom of the queue, people notice, and the anonymous route dies.

Under roughly a hundred employees, a form and a spreadsheet with a named owner will run a suggestion program perfectly well. Above a few hundred, or once you are running campaigns against specific problems, the queue itself becomes the work, and that is what the software is for: routing, deduplication, status tracking, and the reporting that tells you whether you are actually replying inside your window.

What you are buying, in plain terms, is workflow and accountability, not ideas. Vendors in the category price per user and the range is wide. Ideanote, to take one that publishes its numbers openly, verified on August 16, 2026, runs a free tier for up to 10 users with 250 ideas, a Scale tier at $6 per user per month billed yearly or $7 monthly for up to 50 users, and an Ultimate tier starting around $899 per month for 100 users. Several enterprise vendors in the same space publish no pricing at all, which is worth knowing before you build a shortlist. We break the category down on idea management software, the enterprise end on innovation management software, and the numbers side on idea management software pricing.

Buy it when your bottleneck is administrative. Do not buy it expecting participation to rise, because the software does not reply to anybody. It just makes it obvious, in a dashboard, that you have not.

A suggestion program harvests ideas that already exist in people heads. That is genuinely valuable and it has a hard ceiling: the program can only ever collect what somebody already thought of. It does not widen the field, and after the launch burst it will run dry, because the obvious improvements get submitted in the first few months and then the queue thins out.

This is the moment most programs are quietly declared a failure, usually in year two, and it gets blamed on culture or on the tool. Neither is the cause. The backlog thinned because the easy observations were already made, and nothing in the process generates new directions. The program has no divergence step at all.

That is the gap the studio at the top of this page fills. Put the problem your campaign is about into it as one sentence and you get around twenty genuinely different directions in about thirty seconds, each tagged with the angle it came from. Push past the obvious ones with SCAMPER or Six Thinking Hats yourself, then cluster them into themes and score them on impact against effort. Used alongside a suggestion program it does the one thing the program structurally cannot: it puts new options into the queue instead of waiting for them. The reasons programs stall in the first place are worked through in why employee suggestion programs fail, and the scoring method is on idea prioritization.