Market Research Online Community: What It Costs Between Studies

The invoice does not move when the work stops. A standing research community costs the same in a quarter with four studies as in a quarter with none, and nothing in your reporting is built to show that difference. The last study closed, the team moved to the next launch, and the panel sat untouched until somebody remembered it. The platform kept billing. The members stayed members in the technical sense. What they stopped being is a sample, and no dashboard counts that.

What an MROC Is, and Where the Money Starts

The shape is familiar: a private group recruited once and asked things of repeatedly, rather than a fresh sample for every study. The part nobody prices is what happens to that group in the months between studies.

A vendor page answers what a community can do. This one answers what it costs to own through a quarter in which nobody uses it.

The Crossover Test

A community turns a variable cost into a fixed one: a bill per study becomes a bill per year. Whether that trade is good is arithmetic.

The Arithmetic: Cost per Completed Interview, Ad Hoc Versus Standing

Take a 45 to 60 minute B2B interview with a manager or a senior practitioner, self-managed with a paid recruitment channel. Incentive $150, the middle of the published $100 to $200 band for that profile. Recruitment fee $80 at pay-as-you-go rates. A no-show buffer of $42, because a 20 percent no-show rate means recruiting 1.25 people per completed interview, and the wasted slot carries the fee and the scheduling time but not the incentive. Internal labor $86, for 1.5 hours at $57 an hour loaded.

That comes to about $358 per completed interview.

The internal rate is the line worth arguing about: researchers bill at $75 to $150 an hour, and that midpoint adds about $83 per complete, for $441. I kept the lower number, which flatters the community.

Twelve completed interviews, a small study, cost about $4,296 that way. Through an agency, at a planning assumption of $1,000 per complete inside the published $333 to $1,500 band, the same twelve cost about $12,000.

The Break-Even Study Count, Worked

At 400 members, year one of a community costs $108,432, which the table below builds line by line. Divided by $4,296, that is 25 studies a year before the community breaks even against self-managed fieldwork. Divided by $12,000, it is 9 studies against agency-priced work.

Year two is kinder, because founding recruitment is behind you and only 30 percent of members are replaced: 19 studies against self-managed, about 7 against agency.

Scale does not rescue the ratio: 51 studies at 1,000 members, 125 at 3,000. Twenty-five a year is a study every two weeks, which describes a research operation, not a marketing team that keeps a panel on the side. At that rate the community does not pay for itself against fieldwork you would run yourself. It pays against work you buy at agency rates, which is a narrower claim than the one being sold to you.

The Other Half: What a Response Costs

A page that stops at the break-even count is wrong. A community sells responses rather than interviews: 24 activities a year at a 55 percent response rate across 400 members is 5,280 responses. That is $20.54 per response in year one and $15.37 in year two. At 3,000 members it falls to $13.55, then $8.38.

Set that against $358 per completed interview and a response is 17 to 43 times cheaper. A response is a fifteen-minute asynchronous task, not an hour of someone's attention. Give it a quarter of an interview's value and it is still four to eleven times cheaper.

The instrument is cheap per unit of evidence and expensive per decision. Which of those matters depends on how much output you will actually use, and almost nobody forecasts that.

Why Break-Even Arrives Later Than the Pitch Implies

Vendor material sells the community as removing per-study recruitment cost. That is one of four cost lines. The platform, the incentives and the manager are paid in the quarters where no study runs, and founding recruitment, the largest line of year one, never appears in the slide.

What Keeping a Panel Alive Actually Costs

Four lines, three sizes, one year. Every figure is a model with its assumptions printed underneath.

The platform line is the published entry price of an always-on community license, $1,250 a month, which is the floor of a tier that starts at 1,000 participants. I held it flat at 400 members and at 3,000, where the real price is higher, so every community figure here is a lower bound. That is the safe direction to be wrong in.

The management allocations are mine: 0.35 of a full-time person at 400 members, 0.60 at 1,000, one at 3,000. They are anchored on the median wage the US Bureau of Labor Statistics publishes for market research analysts, $37.87 an hour, loaded at 1.4 times to $53.01. They are not taken from an industry survey.

Cost per member falls from $271 to $179, which looks like scale economics and is not: platform and manager are fixed, so they spread. Recruitment is the largest line at every size and rises as a share of year one, from 36 percent to 55 percent.

The Quiet-Quarter Problem

The mechanism has a name, and it comes from a vendor's own guide. Recollective, in a guide updated June 2026, calls it the accordion problem: intense activity during a formal study, silence between studies, and a measurable drop in engagement every time the cycle repeats. Members who are not hearing from you stop answering when you do reach out.

Cost it at the smallest size. Ninety days of silence at 400 members carries $13,398 of fixed cost, $3,750 of platform and $9,648 of management, and saves $3,960 of incentives. You spend roughly $9,400 net to produce nothing, and start the next study on a weaker panel than you ended the last one with.

Cadence is where the published advice contradicts itself. Recollective recommends one to two activities a week. Rival's published case study describes a brand running a couple a month, recording 74 percent response and 94 percent among active members against a platform average of about 55 percent. That is fifty to a hundred activities a year against twenty-four. Both figures are vendor assertions, and the only operator publishing both cadence and response rate runs the lighter one. I would not print a minimum viable cadence as settled, because the only numbers available come from people selling the platform.

Fill the gap with contact that does not pretend to be a study. A check-in on something in the news, a reaction to a product update, a share-back of what the last round changed. None of it needs to be publishable. It needs to remind members they are part of something that values their time consistently, not only when a project is due.

The independent support is old and still the best evidence here. A 2018 practitioner study by MRS with C Space found a feedback loop showing members the impact of their input was up to six times more powerful at sustaining engagement and preventing attrition than financial incentive alone. Read it as a 2018 finding about mechanism, not a current statistic.

When a Standing Community Is the Wrong Instrument

Take the one-off decision first. If you need twelve interviews to choose a name or settle a price, buy them. About $4,300 self-managed, about $12,000 through an agency. A community would leave you carrying a panel for a decision you will not repeat. Where that decision concerns the path a customer takes rather than a number, the artifact it feeds is a customer journey map.

You have not entered the market. You cannot recruit a standing panel of people you cannot describe, and a community built from a segment you do not understand drifts toward whoever answered the advertisement, which you then read as insight.

You need a representative sample. A recruited community is self-selected by construction. If the number has to survive a legal review or a board challenge, you need probability sampling, and good management does not fix that.

Judgment call, and it cuts against the pitch: if you genuinely run four studies a year and buy them yourself, the community costs you more than the fieldwork it replaces.

What the Panel Cannot Tell You

Self-selection is the first limit. The people in your community are the people who joined, which means they care more about your category than your market does. Professional respondents are the second: where an incentive is reliable, a small population learns to treat participation as income, and answers get faster and shallower in ways no quality score catches. The third is structural. Questions about incidence, awareness, market size or non-buyers cannot be answered by a group that opted in. A community tells you what its members think, not who is missing.

How many interviews are enough is a different calculation, worked through in our guide to customer research tools; the interview method itself belongs to voice of customer research. What matters for the budget is where the output lands: a positioning decision ends in a positioning statement, and proof that research pays shows up in the marketing KPIs your team reports, inside the marketing measurement frame.

Where MROC Programs Fail

The pattern is well documented, so you can check for it before you sign.

Under-resourcing was named most often. In a 2018 practitioner study by MRS with C Space, 76 percent of participants said communities became ineffective when under-resourced internally. The worked example is a telecoms team that chose the cheaper self-serve option and found its people scripting surveys all day instead of talking to stakeholders. The saving on paper, in the report's words, felt like a false economy.

Audience drift takes longer to notice. A beauty brand in the same study ran a community for years before the obvious caught up with it: the brand was young and female, the members increasingly were not, and the core had quietly left.

Paying per comment changes the relationship. Civicom, writing in 2022, names the entitlement problem: the more members contribute, the more rewards they feel entitled to, and being obliged to contribute daily pushes them to be less engaged rather than more.

The rest comes down to mandate. Members get asked questions that connect to nothing, participation drops, the team cannot justify the cost, and a community built by chasing a bigger number produces data that looks comprehensive and is shallow. The quietest failure is the most expensive: the research ends when the activity closes instead of when someone acts on it. That is where a panel becomes a cost center with a login.

The Decision Before You Sign

The arithmetic resolves to one question. Are you buying cheaper interviews, or a standing instrument you intend to feed?

If it is cheaper interviews, at brand-side scale you are not buying them. Twenty-five studies a year at 400 members is the self-managed break-even and nine is the agency one. Buy the fieldwork and keep the money.

If it is an instrument, budget the parts that make it one: the cadence, the manager, the recruitment you will redo in year two, and the share-back that keeps members answering. Those lines decide whether the panel is still a sample in month fourteen. The platform fee is the smallest number on the invoice and the easiest to negotiate, which is exactly why it is the one in the deck.