Edit Effort Analysis

The edit effort analysis shows how much work went into a job, based on what the linguist actually changed. The standard word count is calculated before work starts, from where each segment came from (machine translation, a memory match, or nothing). The edit effort analysis is calculated after the job, and sorts the same words into your project's match bands by how much each segment was edited. Use it to compare the predicted effort with the real effort, whether the text came from machine translation, an AI agent, a translation memory, or an earlier step in the workflow.

The analysis is for information only. It does not change the standard word count, costs, quotes, or invoices.

Turning on the edit effort analysis

The analysis is off by default. You turn it on in the word count profile your projects use.

Follow these steps to turn on the edit effort analysis:

1. Go to Settings > Translation Settings > Word counting and open the word count profile.

2. Select Edit details.

3. In the Edit effort analysis section, tick Offer the edit effort analysis on jobs.

4. Click Save. Projects created from this profile after you save will offer the analysis. To apply the change to projects that already use the profile, click Save and spread instead.

The analysis uses the fuzzy match intervals already set in the profile. You don't need to configure any extra bands.

The analysis can only measure jobs created after edits started being recorded per job. Older jobs show Not available for this job (see below).

Viewing the edit effort analysis

A dropdown at the top right of the counts table switches between Standard analysis (the default) and Edit effort analysis. It only changes the counts table.

• Standard projects: open the job and go to the Word count tab. The dropdown is in the table's header bar, to the left of the Excel export button. The Cost tab is not affected.

• Codyt projects: open the project and go to Counts & Cost > Supplier counts and cost. The dropdown is on the right of the action buttons (Export Excel, Clear cost, Set cost, Create quote/invoice). It is not offered on Client counts and cost.

Follow these steps to view the analysis for a job:

1. Open the counts table for your project type, as described above.

2. In the dropdown, select Edit effort analysis.

3. Read the note above the table. It shows when the analysis was calculated, and explains that words are grouped by how much the linguist changed what they received, not by where the text came from. To refresh the figures, click Calculate next to the date.

The table keeps the same columns in the same order. Only the values change:

• The 110%, 100% pre-translated, Machine translated, and Pre-translated fuzzy columns are empty, not zero.

• Those words appear in the match bands and No match instead.

While the edit effort view is on, you can't edit rows, open detail views, or sort and filter by the count columns.

When no figures are shown

The edit effort view shows one of three states:

If a calculation fails, the table keeps showing the last calculated figures with their original date.

Codyt projects

The supplier counts grid shows each job's calculation date or state in its description cell, along with a Calculate button where it applies. Jobs that have been calculated stay visible even when other jobs on the page have no analysis yet. For example, one row can show figures and a date while the row below shows Not yet available and a Calculate button. The footer totals include calculated jobs only, and show how many there are (for example, Calculated jobs on this page: 1 of 2).

The analysis is never used for costs, so in this view the cost columns are empty. Set cost, Clear cost, Create quote/invoice, and row selection are disabled.

How the words are sorted into bands

For each segment, Wordbee compares the text from just before the job's first edit with the text after the job's last edit. Any edits in between are ignored. The difference becomes a similarity score from 0% to 100%:

• Untouched segments score 100%.

• A segment where about a tenth of the text changed scores about 90%.

• A segment that was fully rewritten, or changed by as much as its own length or more, scores 0% and goes to No match.

The score only decides which band a segment goes in. Each segment's source words, characters, and segment count are counted once. Heavy editing moves words to a lower band but never adds more words.

The comparison point is always the text the job received, so the analysis works the same way whatever produced that text. If a job is reassigned part-way, the analysis starts before the first person's edits and ends after the last person's. Both people's work is covered, and nothing is counted twice.

The empty pre-translation columns are expected, and they do not mean data was lost. For example, a machine-translated segment that the linguist left untouched moves from Machine translated to the top match band, because no editing was needed. If you want untouched segments in a separate column, add a 100% interval to the word count profile.

Excluded content

If the word count profile excludes content from the count (non-translatable content, and locked segments when the profile's Locked segments option is ticked), those words also stay in Excluded in the edit effort view. What counts is the decision made when the document's words were counted, not the segment's state when the job ended. Segments a linguist locks while working stay in the analysis, so their work is still measured.

Why the figures can differ from the standard analysis

The edit effort analysis uses the word count stored with each segment. The standard analysis uses the counting rules in the word count profile. Totals can therefore differ, even for a job that covers the whole document. The edit effort view shows this note: "Totals use stored segment counts and may differ from the standard analysis."

Character counts include spaces

The most visible difference is in the Characters column. The edit effort analysis always counts spaces as characters. Whether the standard analysis does depends on the Count algorithm set in the word count profile:

Unless the profile uses MS Word with Count blanks/spaces as characters ticked, the Characters total in the edit effort view is higher than in the standard view, even when the words and segments match.

For example, take the sentence Please send the signed contract back by Friday:

The difference depends on how many words the text contains, so you can't convert one figure into the other with a fixed ratio. Ticking Count blanks/spaces as characters brings the two figures closer, but they can still differ because the standard analysis applies other counting rules (for numbers and symbols, for example).

Other differences

• Net change only: a segment edited several times that ends up exactly as it started scores 100%.

• Overlapping jobs: if another job (a revision, for example) edits a segment between this job's first and last edits, and that change is still there at the end, it counts in this job's figures too.

• Jobs covering part of a document: the analysis counts the exact segments in the job. The standard analysis may share out the whole document's count proportionally, so the totals can differ.

• Very long segments: an edited segment where the text before or after the edit is longer than about 500 characters counts as No match. An unedited segment of any length counts as 100%.

• Lost edit history: splitting segments and some memory operations can remove a segment's edit history. The segment may then look untouched.

• Recounted documents: if a document's words are counted again, the next Calculate uses the new exclusions. Figures already calculated stay as they are until you refresh them.

• Documents counted before this feature was released: no exclusions were recorded for their segments, so all content is treated as not excluded. The Excluded value can therefore differ from the standard analysis.

Learn More

• Job Edit Distance: A single percentage of how much a job's text changed, counting every edit along the way.

• Default word count settings: Set the fuzzy match intervals used for the bands, the count algorithm, and the Count blanks/spaces as characters option.

• Word Counting: How match categories and fuzzy bands drive the standard word count.

• Supplier costs in Codyt projects: How supplier costs are calculated from the standard word count.