Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Research › Troubleshooting

Research Taking Too Long for the Decision

Why research comes too late: no date in the brief, a fuller sequence than needed, recruitment from scratch, late consent work, small studies sent to an agency.
Troubleshootingintermediate17 min readUpdated

First, set the decision date beside the study's plan and find where the weeks went

Start with two dates and a list. The first date is when the decision will really be taken: the meeting, the board paper or the sign-off the research exists to inform. The second is when the study, as planned now, will hand over its findings. If the second falls after the first, the study is late however well it is being run. The research design guide counts research that arrives after the decision among its mistakes.

The list is the study's stages, each with the dates it was meant to start and finish and the dates it did or now will: the request, the agreed brief, the desk work, the design and materials, the consent documents and the data protection review, recruitment, fieldwork, analysis, the report and the readout. Read that list as Cycle Time, the time elapsed from the accepted request to finished work, here the readout, split by stage so that waiting shows up separately from work. A single total only says that the study is slow; the split says where.

Then find the constraint. The theory of constraints treats any system as limited by one step, so effort spent on other steps adds nothing, and in a workflow it looks for the place where work waits longest. In a late study, that is the stage whose real dates ran furthest past its plan, or the stage still ahead whose length nobody can yet promise. Follow the framework's order from there: get more out of that stage before adding capacity, pace the other stages to it, and add capacity only if it still holds the study back. The framework's page also warns that real marketing systems rarely have one obvious constraint, so expect more than one late stage and deal with them in the order they hold the study up.

Two questions for the decision owner complete the picture. How hard is the decision to undo, and which of the study's questions would actually change it? The answers decide how much of the plan the decision needs, and they are the ground for every fix below.

This page starts from a study that is about to miss its decision. A study that reported in time and still changed nothing has a different fault, and the page on findings nobody acts on traces that one from the decision back to the brief.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Nobody can say when the decision will be taken; the study has a topic and an end date set by the methodThe brief's decision field: an action, a date, an owner and the meeting it feedsNo decision date in the brief, so the study was planned without a deadlineFind the decision, its owner and its date, set the standard of evidence from how reversible it is, and book an answer-first readout before it
The plan runs exploration, a representative survey and follow-up interviews in turn, and the decision needs only part of thatEach stage of the plan against the questions the decision turns on, and the evidence the business already holdsThe full explore, measure and explain sequence planned where a compressed sequence would answer the decisionKeep only the stages the decision needs, using interviews already held and a short survey of existing customers, with the limits written down
Fieldwork dates keep slipping while sessions are filled one participant at a timeWhere recruitment began: a new frame, screener and quotas, or people who had already agreed to research contactRecruitment started from scratch for every studyCut quotas to the groups the decision compares, recruit from a frame the business holds with reserves, and start a standing pool
Invitations are drafted but cannot go out, because the data protection owner has only just seen the planWhen the uses, the basis and the retention were first put to whoever owns data protectionConsent and data handling raised with the data protection owner after fieldwork was plannedTake the full list of uses and the drafted documents to the data protection owner now and hold invitations until each use is agreed
A handful of interviews with the business's own customers sits with an agency, and the team's dates depend on the agency'sThe reason for outside help written in the brief: reach, independence, skill, capacity or languageA small qualitative study commissioned from an agency where running it in-house would be fasterRun the interviews in-house, or buy only the part the team cannot do, such as recruitment

Fix, verify and prevent each cause

The brief had no decision date

The research design guide begins every study from a decision, stated in a single sentence that carries an action and a date, and gives three reasons. The decision sets the deadline, since findings that land after it have no value however good they are. It sets the standard of evidence: a reversible decision on a small budget can rest on a few interviews, while an irreversible one that commits a large budget warrants a representative survey and more time. And it makes the sponsor say in advance which finding would change their mind. A brief built on a topic, such as "customer views on the new range", has none of these, so the plan grows to whatever length the method suggests and the end date is set by the work rather than by the decision.

Fix: find the decision the study serves, the person who takes it and the meeting where it will be taken, and write it as the design guide asks. Then set the standard of evidence from how hard the decision is to undo. The framework of one-way and two-way doors divides decisions on that basis: a choice that is easy to reverse is made quickly and watched, while one that is costly to reverse gets research, a test or a pilot first. A decision that can be reversed tolerates the weaker sample the design guide allows, which can free the plan to finish in time. Finally, book the readout before the decision meeting and plan it answer-first, as the guide to reporting research findings describes: the decision, the answer, the team's confidence and its reasons, the main limit and the recommendation, so that the meeting closes with the decision taken, its owners and its dates, or with an honest list of the evidence still missing and the person who will get it. The Readout is where the call is made, and the Decision Log is where it is kept, with the date on which its effect will be checked.

Verify: the decision owner has confirmed the date of the meeting where the decision will be taken, the readout is booked before it, and every stage on the timeline now ends before the readout.

Prevent: on every study, fill the timing field of the research brief template with the decision meeting and its date before anything else, plan each stage backwards from that date, and send back a brief that cannot name it.

The full explore, measure and explain sequence where a compressed one would answer

The qualitative and quantitative comparison sets out a sequence for the two families: explore through secondary evidence plus a small qualitative stage, measure with an instrument built from what the exploration found, then return with a short qualitative stage to explain the surprises in the numbers. That is a strong design for a programme and a long one for a single decision. The same page states that the sequence can compress, with a small team running a few interviews, writing a short survey and sending it to its own customers inside a short window. The design guide reaches the same place from the deadline: it calls timing rarely negotiable, and says that close to a deadline, a properly run round of interviews plus a brief survey of current customers is worth more than a representative study delivered too late.

What compression does not do is save time by switching to a quicker method family. The comparison's own table puts it plainly: interviews cost little to start and take a long time to analyse, and a survey costs a lot to field well and takes little time to analyse once the instrument exists. The time saved comes from dropping stages the decision does not need and from using evidence already held.

Fix: set each stage of the planned sequence against the decision's questions and strike out any stage whose answer could not change the decision by its date. Then check what the business already holds: interviews from earlier studies, open answers from a past survey, sales and support records. Where those cover the exploration, build a short survey of existing customers from them; where the question is about meaning rather than how widespread something is, a set of interviews may answer it on its own, which the comparison accepts when the decision is reversible and the budget small. Write down what the compressed design cannot say, such as anything about people who have never bought, and carry the dropped stages into a later brief in case the decision is revisited.

Verify: every stage left on the plan answers a question the decision turns on; the shortened plan ends before the readout; and its limits are written into the brief, so they reach the first page of the report.

Prevent: state the method in each brief as a preference open to challenge, as the design guide asks, and size the sequence to the decision's date and reversibility rather than to the full programme the topic could support.

Recruitment started from scratch for every study

The customer research programme playbook calls recruitment the slow part of research and keeps it standing for that reason, and its list of mistakes includes restarting recruitment for every study. A study that begins from a blank list has to describe the people it needs, choose a frame, set quotas, write and sign off a Screener, invite, screen, confirm and arrange reserves before the first session, and each of those steps sits on the timeline ahead of fieldwork.

Fix: walk through the steps of the recruiting tutorial and find the one holding the study. Quotas come first: the tutorial keeps them few, because every extra condition slows recruitment and lengthens the screener, so cut them back to the groups the decision actually compares. Draw from a frame the business already holds, such as its customer list, and note who that frame misses. Invite more people than the sessions need and keep reserves from the same quota, as the tutorial does for no-shows, rather than filling an empty session with whoever happens to be free. Do not loosen the screener to fill places sooner; a study that hears the wrong people answers the wrong question on time.

Verify: confirmed participants and reserves fill every quota before the first fieldwork date, and the recruitment log shows who was invited, screened and confirmed.

Prevent: keep a standing pool of customers who have said yes to research invitations, set up as the participant pool tutorial describes: held apart from marketing lists, with a record of what each member accepted and invitations spread so the same people are not drawn for every study. Every study goes on screening its own recruits and logging them; the difference is that it starts from people who have already agreed rather than from nobody.

The participant information and consent tutorial has a study prepare a set of documents before anyone is invited: a list of the uses the study will make of people's answers, the basis and retention for each agreed with whoever owns data protection, participant information in plain words, a separate consent statement for each use, how agreement is captured and where that record is kept, and a final check before the first invitation. It notes that all of this is much easier to settle before anyone is invited than once fieldwork has begun. When the plan, the screener and the invitations are drafted first and the data protection owner sees them last, every question that owner raises holds the invitations back. The design guide lists the same fault among its mistakes: consent, incentives and retention settled during fieldwork instead of in the brief.

Fix: do not shorten, skip or route around the data protection review to recover time; it is not where the time can come from. Take the full list of uses to whoever owns data protection now, with the participant information and the consent wording for every use already drafted, so the owner reviews a complete set rather than a description of one. As general practice, drop any use the decision does not need, such as a recording nobody will watch, because each use adds a statement to agree and a record to keep. Hold the invitations until every use is agreed. The research codes of the MRS and ESOMAR, as the site's guides describe them, ask that taking part be voluntary and informed, and that people learn who is gathering their answers and for what purpose; the invitations go out once those promises can be kept.

Verify: the data protection owner's decision for each use is recorded, the participant information and the consent statements match those decisions, and nothing has reached a participant before that record exists.

Prevent: fill the brief's consent and data handling field when the brief is written, and send the list of uses to the data protection owner on the day the brief goes to the decision owner, so the review runs alongside the design instead of after it.

A small qualitative study sent to an agency

The guide to commissioning outside research repeats the design guide's judgement that, for small qualitative work, doing the study yourself costs less and finishes sooner, while an agency brings access to samples, capacity for fieldwork and distance from what the sponsor hopes to hear. It names the situations that tip the balance towards buying help: reach to people the business cannot contact, independence for a result that will be contested, skill the team lacks, capacity for a study larger or quicker than the team can run beside its other work, and language. The guide also says the argument for buying help weakens when the study is about the business's own customers, involves few people and repeats work the team has done before.

Fix: read the brief against those reasons. If none applies, bring the work in-house: the commissioning guide observes that a few interviews led by a person who knows the product can tell the team more than a bought study it never engages with. If the only gap is reach, buy a part rather than the whole; the guide's table includes recruitment alone, with the design, materials, sessions and analysis staying with the business. Where an agency stays on the study, check that its brief carries the decision and its deadline, and name who signs off each document participants will see before fieldwork, from the screener and invitation to the discussion guide and any stimulus, as the guide asks.

Verify: the brief states which reason for outside help applies, or that none does; the in-house plan, or the part bought, ends before the readout; and an owner is named for each sign-off.

Prevent: decide between in-house and external work when the brief is written, from the commissioning guide's reasons rather than from habit, and keep a short note of the kinds of study the team can run itself.

Prevention: a dated brief, a sequence sized to the decision and recruitment kept warm

The routine that keeps the next study ahead of its decision starts before the brief is written. Search the repository first: the tutorial on building a research repository makes that search the opening move of every brief, so a question already answered is answered from the record instead of being researched again. Then run a desk research pass. Desk Research reads the business's own evidence and published sources first, keeping fieldwork for the questions nobody can yet answer. Both shorten a study by taking questions out of it before any fieldwork is planned.

Take every request through intake. The playbook's Request Intake accepts a request only when it names a decision, an owner and a date, which stops this page's first cause at the door.

Write each brief dated and sized: the decision and its date come first, the method guidance sizes the sequence to the decision's date and reversibility, the consent field goes to the data protection owner the same day, and the choice between in-house and external work rests on a stated reason. Keep recruitment warm through the pool, so that each study draws and screens rather than starting from a blank list.

Before the first session, run the research session readiness checklist. Each gap it catches before the first session is a delay that would otherwise have shown up in front of a participant.

After each study, set its cycle time beside the plan stage by stage and note where the waiting sat. The theory of constraints ends its focusing steps with repeat, because clearing one limit makes another stage the limit, and a rule built around an old constraint can turn into the next one. A team that checks this after every study finds its recurring delay from its own records rather than from a missed decision.

A neighbouring problem has its own page. When a compressed study's interviews come back full of praise and anecdotes, the page on interviews that yield no insight traces the fault through recruitment, questions and synthesis. The marketing research troubleshooting section lists the other problem pages, each built the same way as this one, from a first check to a diagnosis table and a prevention routine.

A sofa maker's study runs late for a range decision (illustrative)

Illustrative example — not a real result

Drimbelqa Sofas recovers a study running late for its range meeting

Drimbelqa Sofas, an invented maker of sofas sold through its own showrooms and website, asks for research on "customer interest in modular sofas". The range meeting that will decide whether to launch a modular range next spring is 9 weeks away, but nobody writes that date into the brief. In week 4 the research owner sets the meeting date beside the plan, stage by stage. The brief sent to an agency asked for the full sequence: 20 exploratory interviews, a representative survey of 1,000 sofa buyers, then 8 follow-up interviews, with a readout weeks after the range meeting. Recruitment for the first interviews started from scratch, with a new screener and quotas for 4 groups, and has confirmed 6 people. The invitations were drafted in week 2, the consent wording reached the data protection owner only in week 3, and nothing has been sent since.

The team writes the decision as "At the range meeting in week 9, decide whether to launch a modular range in 3 showrooms next spring or hold it for a year". A launch in 3 showrooms can be withdrawn, so the decision is treated as reversible and a sample of existing customers is accepted. Drimbelqa already holds 11 interviews with buyers of corner sofas from a study the year before, and they cover what the exploration was meant to find, so recruitment for new interviews stops and the agency's programme is set aside. In week 5 the data protection owner agrees the basis and retention for each use of the answers to a short survey built from those interviews, and that day it goes to 3,000 customers who bought in the past 2 years. It draws 412 responses. The representative survey of non-customers moves to a later brief.

The readout in week 8 opens with the decision, the answer (launch in 3 showrooms), the team's confidence and the main limit: nobody outside the customer base was heard. The range meeting decides to launch in 3 showrooms, and the decision log records the decision, its owner and a check date after the first season. For the next study, Drimbelqa sets up a pool of customers who have opted in to research contact, makes the decision date a required field in its brief and starts a monthly intake.

Common mistakes

  • Planning from the topic. A study with no decision date ends when the method finishes, not when the decision needs it.
  • Treating the full sequence as the only design. Explore, measure and explain suits a programme; a single decision may need part of it.
  • Assuming one method family is quicker. Interviews are slow to analyse and a survey is costly to field well; the saving comes from dropping stages.
  • Building recruitment from nothing each time. The slow part of research, repeated for every study.
  • Loosening the screener to fill sessions. The study finishes on time and hears the wrong people.
  • Sending consent work to the data protection owner last. Every question raised then holds up the invitations.
  • Recovering time from the data protection review. It is not where lost time can come from.
  • Buying a whole study for a few interviews. Outside help pays for reach, independence, skill, capacity or language, not for a handful of conversations with the business's own customers.
  • Fixing a stage that is not the constraint. Time saved anywhere else leaves the study just as late.

Frequently asked questions

The decision meeting is close and the study will not finish in time. What should we do first?

Set the meeting date beside the study's plan, stage by stage, and find the stage that is holding everything else up. Then sit down with the decision owner and agree what the decision needs: how hard it is to undo, and which questions it really turns on. Keep the stages that answer those questions, use evidence the business already holds, and write down what the shortened study cannot say. Book a readout that opens with the answer before the meeting. If the decision is hard to undo and the evidence cannot be gathered properly in time, say so plainly rather than presenting a thin study as a full one.

Can we skip the data protection review to save time?

No. The lawful basis, the retention and the wording participants see belong to whoever owns data protection, and the research codes of the MRS and ESOMAR, as the site's guides describe them, expect taking part to be voluntary and informed. The review is not where lost time can be recovered. What saves time is raising it when the brief is written, with every use of the data listed and the consent statements drafted, so the review runs alongside the design rather than after it.

Is qualitative research faster than a survey?

Not as a rule. Interviews are cheap to start but slow to analyse; a survey is costly to field well but quick to analyse once the questionnaire exists. Time is saved by sizing the study to the decision, not by switching method families: dropping stages the decision does not need, using interviews and records already held, and running a short survey of existing customers where counts are needed. A set of interviews done properly and a brief survey of current customers can be worth more than a representative study that lands after the decision.

Terms used on this page

Frameworks

Sources

  1. Market Research Society — MRS
  2. ESOMAR — ESOMAR

Related