|Cette page en Français||Home page||Services||Past achievements||Contact||Site
|Page d'accueil||Services||Réalisations précédentes||Contact|
|The document contains a certain number of A4 pages||Tp||where 2 ≤ Tp ≤ 200, Tp ≈ 20|
|The document contains a certain number of simple line drawings||Tg||where 0 ≤ Tg ≤ Tp|
|The document contains a certain number of simple tables||Tt||where 0 ≤ Tt ≤ Tp|
Of these pages, diagrams and tables, a certain number (Np, Ng and Nt) are to undergo modification. Typically, Np=Tp, Ng=Tg and Nt=Tt for a new document, but not for the maintenance and update of an existing document suite.
(One point to note is that Np is a count of the number of pages containing at least one modification, even if the modification is to a single character in one paragraph, or a change to a diagram or table that is also counted in the Ng or Nt figure).
The cost of making the changes (largely in the first revision cycle) ought to be proportional to the three parameters, Np, Ng and Nt. Of course, part of the work involved in updating a diagram or table is already accounted for in the value of Np. It turns out that a table costs about the same to edit as the equivalent space occupied by ordinary text, but that a line drawing costs about 3.5 times as much. Thus, there is a factor of 2.5*Ng+0.0*Nt to add to the Np figure.
Next, there are two scaling constants to multiply in to the sum. One is a scaler whose value is 1.0, indicating a normal amount of editing on each page, and the other is a figure, around 0.06, that converts the final value into units of man-days.
Because a technical writer is not constrained in the same way as a translator, the changes to the document can, in places, be quite substantial. In any case, the technical writer will return the edited document to its owner, for appraisal.
If the document owner is not happy with the new state of the document, he/she communicates this to the technical writer, normally in the form of annotations to indicate the subsequent changes that are required. The document goes back to the technical writer, and hence once more round the cycle.
The typical document probably needs just two cycles, and will be "signed-off" after that. However, it is not impossible for a document to need more revision cycles than this. This need not be a reflection on the quality of the job performed by the technical writer; it might be a reflection on the quality of the job performed by the document owner; or then again, it might be a reflection on circumstances that are outside the control of either of them (as described under the heading "Indirect measures").
The cost of multiple revision cycles can be estimated by calculating Tp-(Tg+Tt)/2, and multiplying by the number of revision cycles, Nc, along with a scaler, of around 0.04, to put it in units of man-days. The Tp-(Tg+Tt)/2 expression is based on the assumption that long documents take longer to check than short ones, but that drawings and tables take up page space without requiring so much attention during this simple checking phase, with the average drawing and average table taking up about half a page each.
Finally, there is an offset of 0.5 of a man-day to be added to the total, to allow for such things as the administrative effort of collecting the document from the owner, setting up the necessary files on the computer, and arranging for the publication and archiving of the final document.
As a result of this, the cost of the document, in man-days, can be estimated from the following expression:
|W = 0.06*1.0*( Np + 2.5*Ng ) + 0.04*Nc*( Tp-(Tg+Tt)/2 ) + 0.5|
Here is a small Excel spreadsheet containing this expression (with all cells locked except those for entering the seven numerical parameters). It must be emphasised, again, though that this expression is for guidance only, though it has been confirmed empirically over a large number of cases.
It should be also noted that this gives the estimated cost of the document in man-days of actual expended effort (ignoring time spent doing other things, like getting on with other jobs while waiting for replies to questions about the document). It does not directly give the time that it will take to complete the document.
However, a rough figure can be inferred. Given that the technical writer probably has to schedule this document in with other projects, the technical writer might take at least 2W days to do the W man-days of work. But, there can be many reasons for this delay being longer than average, or shorter. Similarly, the document owner will probably have to schedule the review work of the drafts of this document in with other projects. This might amount to at least 2W days, too. During the process, the technical writer can be in frequent contact (by telephone, or face-to-face) asking for more information, or for ascertaining the urgency of the document relative to the other work that the two people are also currently in the process of doing.
So, the total delay, from beginning to end, can be several weeks; but it is possible, too, to do it all in a single day if the work is urgent, and sufficiently simple and well set out.