What happens at each stage of a doctorate, where the time actually goes, and the decisions that are expensive to reverse.
A doctorate runs through a predictable sequence: finding a question worth asking, establishing what the field cannot yet explain, designing a study that can answer it, collecting and analysing data, and defending the result. The sequence is predictable; the timing is not. Most projects lose time in three places — settling the question, gaining access to participants, and the gap between finishing analysis and knowing what it means.
Doctoral programmes vary by country, discipline and institution, and yours has its own milestones, upgrade points and submission rules that override anything general. What follows is the shape most projects share.
Two things are worth knowing before the detail.
The stages are not sequential in practice. You will be reading literature while collecting data, and rewriting your introduction after your analysis changes what the thesis is about. Presenting the journey as tidy phases makes planning easier and describes nobody’s actual experience.
The early decisions are the expensive ones. A question that is slightly too broad, or a design that cannot support the claim you want, costs very little to fix in month three and can be unfixable in month thirty. Most of the difficulty in a doctorate is concentrated in choices made before any data exists.
Almost everyone starts with a topic that would take three doctorates. Narrowing it is the first real piece of work, and it is harder than it sounds because it feels like giving something up.
The gap. A gap is not an absence in the literature — it is something the field cannot currently explain, predict or decide. “Nobody has studied this here” describes a hole; it does not say why filling it matters. Examiners press on this harder than almost anything else.
The question. A good research question is specific enough that the method it needs is close to obvious. If two entirely different designs would both suit your question, it is still too broad.
What it costs to get wrong. A question that is too broad produces a study that answers nothing convincingly. One that is too narrow produces a study nobody needs. Both are visible early to someone who has examined doctorates, and both are cheap to fix at this stage.
Time is often lost here through reading more in the hope that clarity will arrive. It rarely does. Writing a one-page statement of the question and its consequences, and having someone attack it, is usually faster than another month of reading.
This is where the thesis is really decided, and where the least time is usually spent.
The chain runs from your question to your philosophical assumptions, to an approach, to a design, to sampling, to analysis, to how quality will be judged. Each link has to justify the next, and the whole chain is what the methodology chapter argues.
The decisions that cannot be undone later:
Ethics approval belongs here too, and it takes longer than people expect — weeks or months, with committees pressing on consent, anonymity, data storage and anything involving vulnerable participants. Designing with those questions in view is faster than being sent back.
The stage most likely to derail a timeline, and rarely for methodological reasons.
Access is the binding constraint. Not who you would like to reach — who has actually agreed. Access assumed at proposal stage and never secured is among the most common reasons doctoral projects are redesigned in year two. Where a gatekeeper controls entry, their conditions will shape what you can collect.
Recruitment takes longer than planned. Response rates disappoint, people cancel, and the interesting participants are the busy ones. Building slack into the schedule is more realistic than assuming efficiency.
The invisible work. Transcription is the classic underestimate: an hour of recording commonly takes several hours to transcribe properly, and automated tools still need checking against the audio. For quantitative work, the equivalent is data cleaning, which routinely takes longer than the analysis it precedes.
Pilot first. Two practice interviews or twenty pilot responses reveal which questions are ambiguous while it is still free to change them. Problems found here cost an afternoon; the same problems found after full collection often cannot be fixed at all.
Analysis inherits every limit of the design and can remove none of them. What it can do is answer the question the design was built for, properly.
The stage people are least prepared for is not the analysis itself but what follows it: the distance between having output and knowing what it means. A table of coefficients is not a finding. A list of themes is not a finding. The sentence connecting the pattern to the question — this, given this design and these limits, means this — is work you have to do yourself, and no software produces it.
This is where qualitative projects most often stall, having coded thoroughly and waiting for meaning to emerge from the codes. It will not; moving from codes to claims is a separate activity that has to be started deliberately.
It is also where quantitative projects discover assumption problems, missing data patterns and results that do not support the intended claim. All of those are survivable. What is not survivable is discovering them after the thesis has been written around a conclusion the data does not carry.
Writing is not the last stage. Treating it as one is a common and costly mistake. Draft the methodology while the decisions are fresh; write about findings while you are still close to them. Writing is how you discover what you think, and leaving it all to the end means discovering problems when there is no time to solve them.
The discussion chapter is where most theses are weakest. It is the only chapter that cannot be written by describing what happened, because it requires saying what the findings mean and why that matters. A discussion that restates the findings in different words is the single most common structural failure — recognisable because it cites almost nothing.
The viva rewards preparation of a specific kind. Being able to state your contribution in one sentence, knowing why you chose your design over the obvious alternative, and having an honest account of your weakest point ready. A mock viva with someone who has examined doctorates is the most useful preparation available, because the first time you defend the work aloud should not be the time that counts.
Publication has its own timeline. Turning chapters into papers means restructuring for a different argument and a different length, and peer review takes months. Starting during the doctorate rather than after is the usual advice, with the caveat that it competes for the same time as the thesis.
| Stage | Where time is lost | What prevents it |
|---|---|---|
| Question and gap | Reading further hoping for clarity | Write the question down; have someone attack it |
| Design | Too little time spent here | Review the design before collecting anything |
| Ethics | Amendments and resubmission | Ask the committee what they expect first |
| Access | Assumed rather than secured | Confirm access before the design depends on it |
| Collection | Transcription and data cleaning | Budget for it honestly; pilot early |
| Analysis | The stall between output and findings | Treat interpretation as its own task |
| Writing | Leaving it until the end | Draft each part while the work is fresh |
Broadly: finding an answerable question and establishing the gap it fills; designing a study that can answer it and obtaining ethics approval; collecting data; analysing it and working out what it means; writing up and defending. In practice the stages overlap heavily — you will be reading while collecting and rewriting your introduction after analysis. Your own institution’s milestones and upgrade requirements override any general description.
The two that cause most trouble are settling the question at the start and the gap between finishing analysis and knowing what it means. The first is hard because narrowing feels like losing something; the second because output is not the same as findings, and turning a table or a set of themes into a defensible claim is separate work that no software does for you.
Design, instruments and sampling, essentially. A cross-sectional survey cannot be made to demonstrate causation afterwards by any analysis; a poorly validated measure cannot be repaired statistically; and who you recruited determines what you can claim about anyone else. In sequential mixed methods designs, collecting an anonymous first phase makes purposeful selection for the second impossible. All are cheap to get right beforehand and effectively unfixable after.
Longer than planned, usually for practical rather than methodological reasons. Access is the binding constraint — assumed access that was never secured is a common reason projects get redesigned in year two. Then there is the invisible work: an hour of recorded interview commonly takes several hours to transcribe properly, and data cleaning routinely takes longer than the analysis it precedes.
Much earlier than most people do. Draft the methodology chapter while the decisions are fresh in your mind, and write about findings while you are still close to them. Writing is how you find out what you think, so leaving it to the end means discovering structural problems at the point where there is no time left to fix them.
The cheapest help is the earliest help — a design reviewed before collection, or a question tested before it is committed to. Describe where you have reached and a PhD in your field will tell you honestly what needs attention.
Leave an email or a WhatsApp number — whichever you prefer — and tick how we should reach you. We reach out within 30 minutes.
