The Duplicate Sets page

The Duplicate Sets page shows many AutoMerge duplicate sets on one page, so they can be reviewed and merged quickly. Each set is a box: the winner (rank 1, the record that survives the merge) first, then the other records, which are merged into it. You choose which values survive, then AutoMerge one set or many.

Opening the page

• On a Contact view, use the AutoMerge Duplicate Sets button on the command bar (next to the AutoMerge button). It works on any Contact view, system or personal. Account and Lead have the same button, once it is added to their views.
• With no rows selected, the page shows the duplicate sets of the whole view.
• With rows selected (up to 200), it shows only the sets those records belong to, in the order they were selected, up to 25 sets. The status line says "the sets of the records you selected". Changing a filter or using Reset to view goes back to the whole view.
• Back to view returns to the view the page was opened from.
• The browser tab shows the AutoMerge logo and "AutoMerge Contact Duplicates".

How the page chooses its sets

A duplicate set is the group of active records that share the same AutoMerge tag. A normal Dynamics view filters records one by one, so a view filtered on a member's own field (for example City = Mesa) would list only the members in Mesa: part of a set. The Duplicate Sets page never shows part of a set:

• Only active, tagged records are considered.
• If the view's filter uses only fields that are the same on every member of a set (the tag, the set's precision, the set size, the status), the page reads the sets directly with that filter.
• If the view's filter uses a member's own field (City, Owner, Created On…) or a related table, the page first finds every set in which at least one member matches the view's filter, then reads each of those sets whole, all its members included.

In short: a set is listed when any of its members matches the view. For example, a personal copy of the Dupes view filtered on City = Mesa lists every set that has a member in Mesa, each shown complete, including its members in other cities.

That makes personal views a powerful way to narrow the page: create a personal copy of the Dupes view, add any filter on the records' fields, and open the page from it.

The view's sort order is not used: sets are listed by their precision (highest first), and the records of a set by rank.

The page's own filters work the same way:

Filter	What it does
Sets per page	5, 10, 25, 50, 100, 150 or 200 (the default comes from the settings). Show next N sets at the bottom adds the next page. It is read in the background, so it appears quickly, and the list moves to the first set not yet worked on.
Precision set	From–to, 0 to 100. Your organization can set a lowest value (dspMinimumPrecisionSet); the box then cannot go below it, and its hover text says so.
Set size	From 2 (a set always has at least two records) to any size.
Match type	Lists the sets in which any member was found by that match type.

Reset to view clears the filters and goes back to the view's own sets and columns.

The columns

• The columns are the view's own columns, in the view's order. When useDefaultFields is true, they are the table's fieldList from the AutoMerge settings instead (see Where settings come from).
• Match and Precision always come first. The rank shows in each row, next to the arrow, so it needs no column. On the winner's row, Precision shows the set's precision, marked "(set)". Match shows on the winner's row, and on the other rows only when the set is a hybrid match (found by more than one match type).
• Columns from related tables (for example the company's phone) and computed ones (Full Name) are shown but cannot be chosen. Fields nobody may change (Modified On, Created By…) are shown only.
• The first field of an address block is marked *: choosing its value takes the whole address (all the block's fields) from that record. If a block's first field is not a column, a note at the top offers Add it as the last column.
• Drag a column's edge to make it wider or narrower (double-click the edge for its natural width). Drag a column's name to move it (Match and Precision stay first). Widths, order and added columns are remembered in this browser, per table and view; Reset to view puts the view's columns back.

For how dates and lookups are shown, see Date columns in fieldList and Lookup columns in fieldList.

Reading a set

• Collapsed, a set shows its winner's row with "1 of N" (N is the number of records in the set). The arrow opens it; open, every record shows, ranked 1, 2, 3… Expand all and Collapse all open or close every set (sets start open or closed as the settings say).
• Buttons on each row: Open (opens the record in a new window), Make winner (up arrow), Ignore (red X), and on the winner's row Ignore all (red X marked "All") and AutoMerge (the AutoMerge logo). Your administrator can give these buttons text: the icon then shows followed by the text. Every button explains itself on hover.
• A set that breaks the tagging rules cannot be merged and says why, for example "2 records have rank 1, and a set has only one winner", "no record has rank 1", "it has no rank 2 record", or "its records have different precision set values".
• If an earlier AutoMerge of the set failed, its message shows in red: "Previous AutoMerge attempt failed: date · user · error". Click it to read it whole.

Choosing the values that survive

• In each column, the highlighted value is the one kept on the winner when the set is AutoMerged. To start with, it is the value of the best-ranked record that has one; an address block comes whole from the best-ranked record that has its first field.
• Click another value to keep it instead, or use the keyboard: Tab to a value, then Enter or Space. The value with the keyboard focus has a ring around it.
• Double-click a text value to edit it; it is saved to that record straight away. Select a value and press Delete to empty it (you are asked first).
• Drag a value onto another to copy it. Onto an empty value it is copied at once; onto a value already there, you choose Replace, Swap or Cancel. Hold Ctrl (Option on a Mac) to replace without asking, or Shift to swap (or to move it onto an empty value).
• Values can be dragged within their column, between columns of the same kind (an email into another email field), and within one record. Choices and lookups stay in their own column.
• Values the system requires (such as the Owner) can never be emptied, and Status and Status Reason can be chosen but not changed.
• These changes are written to the records at once; leaving the page does not undo them. After a change, the highlighted values are worked out again. A value you clicked yourself stays highlighted unless it was emptied.

AutoMerging

• One set: use the AutoMerge logo on its winner's row. The set shows "AutoMerging…" with the logo turning, then "✓ AutoMerged N records into <name>", collapsed.
• Several sets: check their boxes (the box in the header checks every set shown) and use AutoMerge selected (N). A confirmation asks first ("AutoMerge (N) sets?"). The sets are merged one after the other; those waiting their turn say "Waiting to be AutoMerged…" and cannot be changed meanwhile. The list glides along so the set being merged stays in view, with the two sets above it.
• A red Stop button appears next to AutoMerge selected while it runs. The set being merged is finished, the rest are not started and stay checked, and a line "Stopped here…" marks where it stopped. Use AutoMerge selected again to continue.

warning

An AutoMerge, like Dynamics' own merge, cannot be undone.

• Before merging, the page reads the set again. If it changed since it was shown, it is shown again for you to check instead of being merged.
• If a merge fails, the values written for it are put back, the reason shows under the set, and the message is kept on the winner ("Previous AutoMerge attempt failed…").
• If the fields chosen to keep do not fit in AutoMerge's list of fields not to merge (the express tag), the merge is refused with a message before anything is written.
• Your administrator can choose whether a confirmation is asked before merging one set, and before merging several (confirmBeforeMergingSingleSet, confirmBeforeMergingMultipleSets).

Make winner, Ignore, Ignore all, and Undo

• Make winner: that record becomes rank 1, and the set is read again in its new order. Precision values are not recalculated (only AutoMerge's analysis does that), so a winner below 100 is normal after Make winner.
• Ignore: that record leaves the set. It is untagged and ignored by future analyses; it can be un-ignored later. The set shows "Ignored <name>. (undo)"; if only one record is left, "Ignored <name>. With only one record left, this is no longer a duplicate set. (undo)".
• Ignore all (on the winner's row): every record of the set is ignored, after a confirmation. The set shows "Ignored all N records of this set. (undo)".
• Undo puts the records back as they were (tags and ranks), as long as the page is open and the set has not been worked on since. If someone else changed the records meanwhile, it says "Can't undo: this set has changed since." and changes nothing.
• A checked set stays checked through Make winner, Ignore and Undo (unless it can no longer be merged).

See also Set Primary and Ignore.

Around the page

• How this page works, next to the title, opens a short summary of this page.
• Light and dark: the page follows the app. The round moon/sun button at the top right switches it for this browser session.
• Row height follows the rowSize setting (sm 28 pixels, md 34 pixels, lg 40 pixels).
• Notes at the top of the page say when something in the settings needs attention, for example a missing table section, or an address block whose first field is not a column.

For which settings the page reads and where they live, see Where settings come from: Other Settings and the widget's settings.

The Potential Duplicates widget

The Potential Duplicates widget on a record's form is the same page showing that record's one set; the record on the form is marked with a bar on its left. Everything under Reading a set, Choosing the values that survive, AutoMerging (one set) and Make winner, Ignore, Ignore all, and Undo applies there too. For the widget's own settings, see Customizing the subgrid with the Custom Parameters JSON.