Almost nothing else about length is published
That is the fact worth taking from it. Where a length rule exists, somebody had to write it down for an operational reason: an award, a competition, a journal's word limit, a contract's stated manuscript length. Everywhere else the figures that circulate are conventions rather than rules, and the bodies that could publish them do not.
An author asking how long a book should be is usually asking three different questions at once. What length will a publisher accept? What length does this argument need? And what will this length cost to produce?
Only the first is a matter of somebody else's rule, and the answer to it lives in the contract or the submission guidelines rather than in general advice. The other two are the author's to settle.
Length has a price, and the price is arithmetic
The third question has a published answer. Editing is commonly priced by the word, and the Editorial Freelancers Association's rate chart reports per-word figures for every level of work.2 A manuscript 30,000 words over its intended length costs more to copyedit than one at length, by a factor anyone can calculate before deciding anything.
Production follows the same arithmetic. More words mean more pages, more paper, and a higher unit cost, and those figures belong to the publisher rather than to the author. What an author can do is know that a length decision is also a budget decision, and make it deliberately.
Cutting and reordering are different operations
A book that is too long and a book that feels slow have different problems, and only one of them is solved by cutting. Telling them apart is the first piece of work.
Editors Canada's structural editing standards cover both. They ask the editor to reorganize material to achieve a coherent structure and sequence, with a logical progression of ideas and a narrative or expository flow. They separately ask the editor to revise, cut, or expand text, or suggest changes, to meet length requirements.3
Those are two standards because they are two jobs. A manuscript 30,000 words over a stated limit has an arithmetic problem. A manuscript of the right length that loses its reader in the middle third has an organization problem, and cutting will not fix the order.
The diagnosis matters commercially as well as editorially. An author who cuts when the problem was sequence has spent a month and produced a shorter book with the same fault.
What an editor can report about length
Four things, and none of them is an opinion about how long the book ought to be.
The first is the current count, by chapter as well as in total. The second is the distribution: a book whose chapters run from 2,000 to 14,000 words has told the reader something about its own priorities, and usually not deliberately.
The third is where material repeats. Two chapters making the same case are a length problem with a structural cause. Editors Canada's standards ask an editor to identify and either recommend or make deletions and additions, so that the material is suitable.3
The fourth is where the count and the text disagree. A stated chapter count, a table of contents, and the chapters actually present should agree, and on a long manuscript they often do not. Checking that is copy editing work, covering correctness, consistency, and completeness.4
What stays with the author
Whether a book should be shorter is not an editorial judgment, and neither is which material earns its place. Those decisions belong to the author, and where a publisher has set a limit they belong to the contract.
Everything a client sends is treated in confidence, including unpublished manuscripts, outlines, and correspondence with a press or an agent. EditFast quotes novel and book-length work on the whole manuscript, so the figure reflects the length that exists rather than the length that was hoped for.
An author who wants one number before doing anything else can take the count by chapter. It costs ten minutes and settles more arguments than any general figure about what a book should weigh.