Thesis Writing Tips

There are lots of tips, guidelines and HOWTOs out there.

This page contains an additional summary of most asked aspects and Do/Don'ts I recognized during last theses supervision.

Length of your chapters

There is no hard rule, because your topic is new and unique. Thus, your structure and balance of the parts is unique.

Ignoring the specific needs of your topic, a rough rule of thumb is

  • 30% for state of the art and related work (showing you know the field and knowledge you build on)
  • 30% for your model of the new proposed solution (architecture, structures, communication etc)
  • 30% for your implementation and evaluation (show some interesting code and proof that it is good)
  • 10% for intro, motivation, problem and conclusion, outlook

In your model part: Think big without restrictions of time and man power. Propose a holistic, complete, and perfect solution concept. The architecture, needed parts, data structure etc. should be complete in all relevant aspects to solve the problem of your thesis.

In your implementation part: Mostly, you only implement some of the model parts. Argument consisely for the reader, which parts you chose for prototype implementation, why these are the most relevant ones to proof the functionality. It should be a selection of components of your model that are at least represent a red thread (a thin connected line) of your components. That eases your evaluation and proof that the concept you propose with your model is working.

DOs and DON'Ts


  • Give each picture, graph, chart, code listing and table a number and description. Reference these numbers in the text (e.g. "..for a visual presentation of the results see Fig. 3").
  • Colored diagrams and info graphics should still be comprehensable in black/white printing. Thus, in your legend never refer to red and green lines or bars, but add an additional pattern or difference in lines. Then a reader can recognize the differences even when the page is printed in b/w.
  • Only use quotation marks (" ") for word-for-word quotes. If you quoted words from a source then directly after the closing quotation mark (") the reference has to be given with page number.
  • Use italics (in LaTeX \emph{}) to highlight terms or phrases instead of using quotation marks (" ").
  • Add references to basically all statements or facts you write that you have not found yourself but have been written somewhere.
  • Give reference at the end of the sentence. Only if the author/ref is part of the sentence put it inside (e.g. "As Smith et. al (2015) have shown, ...")
  • If a whole paragraph has one reference put it at the end of the paragraph outside the full-stop (.). Thus make use of paragraphs to structure your arguments and thoughts.
  • For all data you present in form of mean values (M) or bar charts: always add confidence intervals and name ther standard deviation in text, too. E.g. "On average users gave 5.2 of 6 points (SD=0.44)." Only with a deviation metric such values are meaningful (e.g. mean of 3 and 3 is 3, and of 1and 5 it is as well 3, so what does it tell us about the data if we only know M=3?)
  • If you compare any A and B in an evaluation, then calculate the significance of the difference in values. It is of low quality to report only mean values., e.g. use t-tests.


  • Don't use vague numerical expressions like many, some, recent years, some time, few, most, ("viele, einige, verschiedene, in letzten Jahren, etwas Zeit, wenige, die meisten). It always gives the impression that you don't know the exact number or amount. And the reader is lost. What is few? 2? or 2000? So always give specific figures, percentages, or at least write "..more than y%" or "less than x" and give a reference (source) for it from your literature or own testing.
  • Don't use colloquial language (Umgangssprache). This means no "natürlich", "sollte klar sein", "liegt auf der Hand", "..die Performance ist unglaublich..", "spricht für sich selbst"
  • Don't use irony or rethorical questions
  • Don't use the word "important". It implies unspecific unimportant aspects. Better write: considerable, relevant, unexpected, noticable, ..
  • Don't use first- or second personal pronouns, like I, we, you. Instead formulate indirect and passive sentences. (e.g. "The author of the thesis", "After doing thisthat it could be seen that...", "Based on these facts, it can be concluded...", "The source was transformed from y to x..")
  • Don't put links to tools or platforms into the literature / references. This means, URLs to, or etc. should be a footnote (with date of access). Add to your reference section only text-based references. This can include URLs to texts like tech docu, howtos, blog posts, but (as said) never any software or tool.
  • Never cite any Wikipedia article or other wikis. Look what they name as reference. Aim for highly reliable, trust-worthy sources  (Journals, good conferences and sometimes books).