The ArcGIS Cheat Sheet Pack

The ArcGIS Cheat Sheet Pack

Ten dense quick-reference cards distilled from the ArcGIS Compendium. Find any answer in under ten seconds; chapter references point into the eight volumes.


Map Viewer at a Glance

One-page reference for the ArcGIS Online Map Viewer. Left toolbar = what is in the map (Contents). Right toolbar = what you do to the selected layer or the map (Settings). Full treatment: Compendium Chapter 6 (Map Viewer Complete Reference).

Left toolbar (Contents)

Entry One-line purpose
Layers The layer list: drag to reorder, eye icon to toggle visibility, ... menu for per-layer actions. The Add button at the top of this panel brings in new layers.
Tables Standalone (non-spatial) tables added to the map; view and filter records.
Basemap Swap the basemap from your org's gallery.
Charts Build charts from layer attributes; charts save with the map.
Legend Read-only symbol key for visible layers. Nothing is configurable here — change symbols in Styles.
Bookmarks Saved extents you can add, rename, and reorder. Saved with the map.
Save and open Save, Save as, Open map. The only place your work gets written.
Map properties Map-level options such as the map background color.
Share map Set the map item's sharing level; prompts you about under-shared layers.
Print Export a printable layout of the current view.
Create app Launch an app builder (Instant Apps, Dashboards, Experience Builder, StoryMaps — Compendium Chapters 26-29) from this map.

Right toolbar (Settings)

Most panes act on the currently selected layer in the Layers panel — select first, then configure. Entries vary with layer type and your org privileges.

Entry One-line purpose
Properties Visibility range, transparency, blend modes, and a symbology summary for the selected layer.
Styles Smart mapping: choose attributes, choose a drawing style, adjust symbols (Compendium Chapter 7).
Filter Attribute expressions that limit which features draw.
Effects Visual effects (drop shadow, bloom, etc.), optionally split by a filter expression.
Aggregation Clustering and binning for point layers, including aggregate pop-ups.
Pop-ups Enable, disable, and compose pop-up content blocks (Compendium Chapter 9).
Fields Map-level overrides for field aliases, decimal places, and date formats.
Labels Label classes, expressions (Arcade — Compendium Chapter 8), style, and visible range.
Forms Build the edit form used by Map Viewer editing and Field Maps (Compendium Chapter 30).
Edit Create and modify features on editable layers. Edits commit to the layer immediately.
Analysis Run spatial analysis tools; requires privileges and most tools consume credits (Compendium Chapters 16-20).
Search Find places and addresses; also accepts typed coordinates.
Measurement Distance and area; change units inside the tool's dropdown.
Directions Point-to-point routing; requires network analysis privileges and consumes credits.
Sketch Draw graphics on a sketch layer; saved with the map, not to any feature layer.

Save and share semantics

Where commonly-hunted things hide

You want Where it is
Rename a layer Layers > ... > Rename — changes the map's label only, not the layer item
Attribute table Layers > ... > Show table
Zoom to a layer Layers > ... > Zoom to
Transparency / visible range Select the layer, then the Properties pane
Turn a pop-up off Select the layer, Pop-ups pane, toggle at the top
Field alias or decimals Select the layer, Fields pane (map-level only)
Cluster on/off Select a point layer, Aggregation pane
Rotate the map Right-click and drag; reset north with the compass button
Refresh interval The layer's options menu in the Layers panel
Undo a config change There is no undo — reopen the map without saving

Five settings that live on the item page, not in the map

Open via Content > (item) > Settings unless noted.

Setting Why it matters
Delete protection Prevents the map (or layer) from being deleted accidentally. Turn it on for anything in production.
Offline use Whether the map can be taken offline in Field Maps. Not configurable inside Map Viewer.
Extent override Sets the item's default extent independently of what was saved in the viewer.
Item metadata Title, summary, description, thumbnail, terms of use — on the Overview tab. This is what search, galleries, and many apps display.
Layer capabilities Editing, sync, and export permissions live on each layer's item page, never in the map (Compendium Chapter 10).

Watch out: Sharing the map without sharing its layers. The map looks perfect to you because you own everything; everyone else opens a blank or partially broken map, and nobody tells you. When the share dialog warns that layers are shared more narrowly than the map, deal with it then — do not dismiss it.


Arcade Quick Reference

Compact reference for ArcGIS Arcade. The full tutorial is Compendium Chapter 8 (Arcade from Zero to Fluent); pop-ups and labels are Chapter 9, styling expressions Chapter 7, dashboards Chapter 28, forms Chapter 30, attribute rules Chapter 14.

Syntax essentials

Construct Syntax Notes
Variable var x = 10 Semicolons optional
Field value $feature.STATUS Field name, not alias
Field with spaces $feature["Install Date"] Bracket form
Comment // line or /* block */
Two-way choice IIf(cond, a, b) Returns a or b
Multi-branch When(c1, r1, c2, r2, default) First true condition wins
Block logic if (c) { } else if (c2) { } else { } Always use braces around bodies
Loop for (var i in arr) { arr[i] } i is the index
Function function f(x) { return x * 2 } Explicit return inside
Comparison == != > >= < <=
Logic && || ! SQL strings inside Filter() use AND/OR instead
Join text "A" + "B" or Concatenate([a,b], ", ")
Template text `Total: ${x}` Backticks, current versions
Null guards IsEmpty(v), DefaultValue(v, "n/a") Use constantly
Result Last evaluated statement is returned Or explicit return

Most-used functions by task

Task Functions
Text Upper Lower Proper Left Right Mid Find Replace Split Trim Count (length)
Formatting Text(value, "#,###") for numbers; Text(date, "MMM D, Y") for dates
Math Round Floor Ceil Abs Min Max Pow Sqrt Number (parse)
Date Now Today Date DateDiff DateAdd Year Month Day Weekday
Geometry Geometry($feature) AreaGeodetic LengthGeodetic Distance Buffer Centroid Intersects Contains Within
FeatureSet FeatureSetByName FeatureSetByPortalItem FeatureSetByRelationshipName Filter First Count OrderBy Top Sum Mean GroupBy Distinct
Domains / misc DomainName Decode Console (debug) Push (append to array)

Where Arcade runs (profiles)

Where you write it Profile FeatureSet access Must return
Pop-ups Popup Yes Text, number, or rich-content dictionary
Styling attribute expressions Visualization No Number or category text the style expects
Label expressions Labeling No Text
Smart forms (calculated, constraint, visibility) Form Yes in calculations; limited elsewhere Field-type value; constraint/visibility return true/false
Calculate field Calculation Yes Value matching the field type
Attribute rules (Pro/Enterprise) Attribute rule Yes, plus $datastore Value, or dictionary describing edits
Dashboard data expressions Data expression Yes (portal items) A FeatureSet

An expression that compiles in one profile can fail in another — test in the profile you will use.

Watch out: Calculate field on a hosted feature layer writes values immediately and permanently. There is no undo. A wrong expression silently overwrites every row it touches. Prove the expression in a pop-up or on a copy of the layer first, and calculate against a filtered selection before running it on everything.

Gotchas checklist

15 copy-paste snippets

1. Null-safe display — use when a field may be empty and you do not want "null" showing.

DefaultValue($feature.OWNER, "Not recorded")

2. Combine fields, skipping blanks — use when building one line from sometimes-empty parts.

var parts = [$feature.STREET, $feature.CITY, $feature.STATE];
var out = [];
for (var i in parts) {
  if (!IsEmpty(parts[i])) { Push(out, parts[i]) }
}
Concatenate(out, ", ")

3. Thousands separators — use when raw numbers are unreadable.

Text($feature.POP2020, "#,###")

4. Readable date — use when showing a date field in a pop-up or label.

Text($feature.INSTALLED, "MMM D, Y")

5. Days since — use when flagging overdue inspections or stale records.

Round(DateDiff(Now(), $feature.LAST_INSP, "days"), 0)

6. Bucket a number — use when classifying a value for styling or display.

When($feature.SCORE >= 80, "Good", $feature.SCORE >= 50, "Fair", "Poor")

7. Translate codes without a domain — use when the layer stores codes but no domain exists.

Decode($feature.STATUS, 1, "Open", 2, "Closed", "Unknown")

8. Show the domain description — use when a coded value domain exists and you want its label.

DomainName($feature, "STATUS")

9. Safe percentage — use when dividing fields that can be zero or null.

IIf($feature.TOTAL > 0, Round($feature.PART / $feature.TOTAL * 100, 1) + "%", "n/a")

10. Acres from geometry — use when there is no area field or it is stale.

Round(AreaGeodetic($feature, "acres"), 2)

11. Value from an intersecting layer — use when a pop-up needs another layer's attribute without a join. Use if/else, not IIf, so the attribute is never read from a null feature.

var zones = FeatureSetByName($map, "Zoning", ["ZONE_CODE"]);
var z = First(Intersects(zones, $feature));
if (IsEmpty(z)) { return "No zone" }
return z.ZONE_CODE

12. Count related records — use when a parent pop-up should show how many children exist.

Count(FeatureSetByRelationshipName($feature, "Inspections"))

13. Sum a field across related records — use when totaling child values in the parent.

var insp = FeatureSetByRelationshipName($feature, "Inspections", ["COST"]);
Sum(insp, "COST")

14. Two-line label — use when a label needs a name plus a stat.

$feature.NAME + TextFormatting.NewLine + Text($feature.CAPACITY, "#,###")

15. Dashboard data expression: grouped counts — use when a dashboard element needs statistics the layer does not expose. See Compendium Chapter 28 (Dashboards).

var fs = FeatureSetByPortalItem(Portal("https://www.arcgis.com"),
         "<item-id>", 0, ["STATUS"], false);
GroupBy(fs, "STATUS",
  {name: "total", expression: "1", statistic: "COUNT"})

Analysis Tool Picker

Find your question, run the tool, apply the sanity check before trusting the output. Credit flags are qualitative — rates change; see Compendium Chapter 2 (The ArcGIS Ecosystem). Legend: None = Pro geoprocessing on local data; Std = standard Online analysis credits, scales with feature count; High = premium service (routing, GeoEnrichment, raster/elevation), scales fast. Tool names show Online first, then Pro where they differ.

Proximity

Depth: Compendium Chapter 16 (Proximity and Overlay).

You want to know Tool (Online / Pro) Key inputs Credits Sanity check
What lies within X distance of these features? Create Buffers / Buffer (or Pairwise Buffer) Layer, distance + units, dissolve on/off Std / None Measure one buffer edge with the measure tool; width matches your distance
Which feature is closest to each point, straight-line? Find Closest, straight-line option (Find Nearest in Map Viewer Classic) / Near or Generate Near Table Two layers, max distance, match count Std / None Spot-check one pair visually on the map
Which features are near others — selection only, no new layer? (Pro) Select By Location, on the Map ribbon Relationship, distance, selecting layer None Zoom to selection; count is plausible

Overlay and combine

You want to know Tool (Online / Pro) Key inputs Credits Sanity check
Where do two layers overlap? Overlay Layers, Intersect mode / Intersect (or Pairwise Intersect) Two layers, output geometry type Std / None Output area never exceeds the smaller input
What part of A is outside B? Overlay Layers, Erase mode / Erase (or Pairwise Erase) Input layer, erase layer Std / None Erased + intersected pieces roughly rebuild the original
All pieces of A and B, attributes from both? Overlay Layers, Union mode / Union Two polygon layers Std / None Output attribute table carries fields from both inputs
Trim data to a study area? Overlay Layers, Intersect with boundary / Clip (or Pairwise Clip) Input layer, boundary polygon Std / None Nothing renders outside the boundary
Stack same-schema layers into one? Merge Layers / Merge or Append Layers with matching fields Std / None Row count equals the sum of inputs

Summarize and join

You want to know Tool (Online / Pro) Key inputs Credits Sanity check
How many points fall in each polygon? Aggregate Points / Summarize Within or Spatial Join Points, polygons, stats fields Std / None With non-overlapping polygons, sum of counts ≤ total points (points outside all polygons drop)
Sum or average a field inside each polygon? Summarize Within (same name in both) Polygons, layer to summarize, statistic Std / None Grand total matches the source table's total
What surrounds each feature within a drive time? Summarize Nearby (same name in both; uses the routing service) Features, travel mode + time, stats High Urban vs rural features show very different totals
Attach fields by a shared ID? Join Features (attribute option) / Add Join or Join Field Key field each side, join type Std / None Check the unmatched-record count before using results
Attach attributes by location? Join Features (spatial option) / Spatial Join Spatial relationship, one-to-one vs one-to-many Std / None One-to-many multiplies rows — compare counts before and after
Add demographics around my features? Enrich Layer / Enrich (GeoEnrichment service behind both) Variables, buffer or drive-time area High Check one known area's population against a census figure

Patterns and statistics

Depth: Compendium Chapter 17 (Statistical and Pattern Analysis).

You want to know Tool (Online / Pro) Key inputs Credits Sanity check
Significant clusters of high/low values? Find Hot Spots / Hot Spot Analysis (Getis-Ord Gi*) Analysis field or point counts, aggregation scale Std / None Hot spots persist at a different aggregation scale
Features unlike their neighbors? Find Outliers / Cluster and Outlier Analysis Analysis field Std / None Inspect one flagged outlier — the contrast should be obvious
Where are points densest? Calculate Density / Kernel Density (Spatial Analyst) Points, search radius, cell size Std / None Peaks sit where you visibly see the most points
Natural groupings of points, no polygons? Find Point Clusters / Density-based Clustering Min features per cluster, search distance Std / None Expect unassigned "noise" points — zero noise suggests bad parameters
Predict values between sample points? Interpolate Points / IDW or Kriging (Spatial Analyst) Sample points, value field Std–High / None IDW output range stays within the measured min/max

Network

Depth: Compendium Chapter 18 (Network Analysis). All of these use street-network routing, not straight lines.

You want to know Tool (Online / Pro) Key inputs Credits Sanity check
Area reachable within X minutes' drive? Generate Travel Areas / Service Area (Network Analyst) Facilities, travel mode, time/distance breaks High Shape hugs the roads — a neat circle means you got straight-line by mistake
Best route through several stops? Plan Routes / Route Stops, travel mode, optimize order on/off High Compare one route's time against a mapping app you trust
Closest facility by travel time? Find Closest, travel-mode option / Closest Facility Incidents, facilities, cutoff High Nearest-by-road can differ from nearest-by-air — verify one case

Raster and zonal

Depth: Compendium Chapter 19 (Terrain and Raster Analysis) and Chapter 15 (Imagery and Rasters).

You want to know Tool (Online / Pro) Key inputs Credits Sanity check
Average raster value per polygon (rainfall per district)? Zonal Statistics as Table (Pro, Spatial Analyst; Online raster analysis needs imagery privileges) Zone polygons, value raster, statistic High / None Every zone mean falls inside the raster's min/max
Slope, aspect, hillshade from elevation? Slope, Aspect, Hillshade (Pro surface tools); in Online, Living Atlas terrain layers cover most display needs, and computing your own requires raster analysis with imagery privileges Elevation raster or terrain service High / None Steepest slopes align with tight contour spacing
What is visible from a point? Create Viewshed (ready-to-use elevation service) / Viewshed (Spatial Analyst) Observer points, height, distance High / None Ridgelines block visibility behind them

Naming traps

Before you run anything

Watch out: The costliest mistake here is running a credit-hungry tool (Enrich Layer, Generate Travel Areas, Interpolate Points, raster analysis) against an entire large layer when you meant a filtered subset — with the extent toggle set wrong. Credits are spent even when the result is useless. Filter first, confirm the extent toggle, and read the credit estimate before every run.


Projections Quick Card

One-page lookup for choosing, checking, and fixing coordinate systems. Full treatment: Compendium Chapter 3 (Coordinate Systems and Projections).

Pick a system by task

Your task Use this Why
Web map display Web Mercator (EPSG:3857) Matches every standard basemap and tile service; tiles align, pan/zoom is fast
Comparing areas or densities An equal-area projection (Albers for continental US, Equal Earth or Mollweide for world) Only equal-area projections preserve relative size; Web Mercator wildly inflates high latitudes
Regional engineering, surveying, utilities (US) State Plane zone for your area Designed for very low distortion inside its zone; matches surveyor and CAD data
Regional work outside the US, or multi-state corridors UTM zone for your area Worldwide zone grid, metric units, low distortion within one zone
Global data exchange, storage, GPS coordinates WGS84 geographic (EPSG:4326) Universal lingua franca; every system reads lat/long on WGS84
Navigation with constant compass bearings Mercator Straight rhumb lines; its one legitimate specialty
Polar mapping A polar stereographic projection Web Mercator cannot even show the poles

Decision shortcut: display on the web = Web Mercator, measure or analyze = a local projected system, store and share = WGS84, compare sizes = equal-area.

The four systems you will actually meet

System Kind Units Built for Do not use for
WGS84 (EPSG:4326) Geographic (datum + lat/long, not a projection) Decimal degrees GPS output, data storage and exchange, API defaults Planar measurement of any kind; a "degree" is not a fixed ground distance
Web Mercator (EPSG:3857) Projected, conformal-ish Meters (nominal, badly distorted with latitude) Web basemaps, tiled services, ArcGIS Online display Measuring distance or area, comparing region sizes, anything legal or engineering
UTM (60 north/south zone pairs, each 6 degrees wide) Projected, transverse Mercator Meters Regional analysis and fieldwork anywhere on Earth Data spanning multiple zones without reprojecting to one
State Plane (US only, per-state zones) Projected, zone-specific Feet or meters — always check which County and state engineering, parcels, public works Anywhere outside the designed zone; assuming units without checking

Checklist before any analysis:

Geodesic vs planar measurement

Planar Geodesic
How it works Straight-line math on the flattened map Computed on the curved ellipsoid
Correct when Data is in a projection designed for that area and that measure Always, in any coordinate system
Wrong when Data is in Web Mercator or WGS84, or spans a large extent Rarely; slightly slower

The rule: if the data is in Web Mercator or WGS84, or covers more than one zone's worth of ground, measure geodesically. Map Viewer's measure tools work geodesically. In Pro, pick the geodesic option in the Measure tool, in right-click field > Calculate Geometry, and in tools like Buffer (choose the geodesic method). Arcade has geodesic geometry functions for pop-ups and calculated fields (Compendium Chapter 8).

Buffers deserve special mention: a planar buffer on Web Mercator data is not a circle on the ground and its size error grows with latitude. Use geodesic buffers (Compendium Chapter 16).

Project vs Define Projection

Two tools in Data Management Tools > Projections and Transformations that sound alike and do opposite things:

Tool What it changes Use when
Project The coordinates themselves — data is mathematically converted to the new system and stays in the right place You want the data in a different coordinate system
Define Projection Only the metadata label — coordinates are untouched The label is missing or you know for certain it is wrong (e.g., "Unknown coordinate system")

Watch out: Running Define Projection to "fix" a layer that draws in the wrong place is the single most damaging projection mistake. It relabels the coordinates without converting them, so the data lands anywhere from a few hundred meters off to a different continent — and the file now lies about what it is, which poisons every later workflow. Define Projection is for correcting a wrong or missing label only. If the data needs to be in a different system, use Project.

Fast diagnostics

Symptom Likely cause Fix
Layer draws in the ocean near the equator/prime meridian (0,0) Coordinates are lat/long degrees but labeled as a projected system, or the label is missing Define Projection with the true source system, then Project if needed
Layer is offset by a consistent amount (tens to hundreds of meters) Datum mismatch — usually NAD27 vs NAD83/WGS84 with no transformation applied Reproject with the correct geographic transformation
Areas or distances look inflated toward the poles Measuring planar in Web Mercator Switch to geodesic measurement, or project to a local system
Two "identical" layers do not line up at high zoom Different datums or a sloppy define somewhere upstream Check each layer's coordinate system in its properties; fix at the source
Web layer publishes fine but sits in the wrong spot Wrong define before publishing Fix the source data and republish (Compendium Chapter 10)

Defaults worth setting in Pro


Layer and Format Picker

Use this sheet when you are staring at "Add layer" or "New item" and need to know what a layer type can do, or what a file will turn into after upload. Deep treatment of each layer type is in Compendium Chapter 10 (Hosted Feature Layers), Chapter 15 (Imagery and Rasters), and Chapter 35 (ArcGIS Enterprise).

Layer type matrix

Layer type Clickable (pop-ups)? Editable? Restylable? Best for
Feature Yes Yes, when hosted with editing enabled Yes, fully (smart mapping, Arcade) The default. Anything you query, filter, edit, analyze, or style yourself
Vector tile No attribute pop-ups No Yes — colors, fonts, visibility via the style editor Basemaps; huge datasets that must draw fast and never change
Tile (raster cache) No No No — pixels are pre-rendered Static reference maps, cached scans, legacy basemaps
Imagery Yes — pixel values and attributes via identify No Rendering only (stretch, band combos, colormaps) Satellite and aerial imagery, elevation, analysis rasters — Chapter 15
Map image Yes (queries go to the server) Not directly; edits go through a companion feature endpoint Limited; the server draws it Many-layer maps published from ArcGIS Enterprise — Chapter 35
Scene Yes for point and 3D object scenes (attributes are cached with the layer); integrated mesh and point cloud carry little or none No; editing goes through an associated feature layer where one exists Limited (3D symbology) 3D buildings, meshes, point clouds in Scene Viewer
Table No map presence at all Yes, when hosted n/a Related records, joins, standalone lists — Chapter 12
Group Inherits from children n/a Children styled individually Organizing the layer list; one visibility toggle for a set
Media No Placement only (position, transparency) No Georeferenced images: floor plans, scanned historical maps

Quick rules of thumb:

Import format table

All of these upload from Content > New item, or straight into a map from the Add layer menu in Map Viewer. Full walkthroughs are in Compendium Chapter 11 (Creating Data).

Format Prep before upload What it becomes Gotchas
CSV / Excel One header row; latitude/longitude columns, or clean single-purpose address columns; no merged cells or formulas in headers Hosted feature layer (points) or hosted table Address geocoding consumes credits — Chapter 2. Lat/lon columns cost nothing. Check field types after upload; ZIPs love to become numbers
Shapefile Zip the whole set together — .shp, .shx, .dbf, and .prj at minimum, no nested folder Hosted feature layer Missing .prj = data lands in the wrong place (Chapter 3). Field names truncate at ten characters; long text and dates can be lossy
GeoJSON Coordinates should be WGS84 (the spec requires it); one file Hosted feature layer No field aliases or domains come through; very large files upload slowly
KML / KMZ None — upload as-is A linked KML layer with styling baked in Effectively read-only: not editable, not restylable, weak pop-ups. Convert to a feature layer if you need to do real work with it
GPX Export straight from the device or app Hosted feature layer (waypoints and tracks) Attributes are minimal; timestamps usually survive, custom fields usually do not
File geodatabase Zip the .gdb folder itself Hosted feature layer(s) — keeps domains, aliases, long field names, multiple feature classes The highest-fidelity path from ArcGIS Pro. Prefer it over shapefile every time — Chapters 11 and 12

Format decision in one line: have Pro → file geodatabase; have coordinates in a spreadsheet → CSV; handed a shapefile → zip it; handed KML → convert it.

Hosted layer vs file dropped into a map

Small files can be added straight into a single map instead of becoming a hosted layer item. The convenience is real; so is the trap.

Hosted feature layer File added directly to one map
Where the data lives Its own item in Content Embedded inside that one map
Reusable across maps/apps Yes — one source, many maps No — re-upload per map, copies drift apart
Editable later Yes, with editor tracking and views No — the embedded copy is a static snapshot; to change it, re-upload
Size and performance Scales; indexed and queryable — Chapter 10 Small files only; the whole thing loads with the map
Shareable / securable on its own Yes (sharing levels, view layers) No — rides on the map's sharing
Cleanup burden One item to manage Hidden data in every map that used the shortcut

Pick hosted when any of these is true:

Pick in-map only when all of these are true:

Watch out: Dropping the same file into several maps creates independent embedded copies. A fix or restyle in one map changes nothing anywhere else, and months later nobody can say which copy is current. If the data matters at all, publish it once as a hosted feature layer and add that layer everywhere.

Ten-second answers

Question Answer
Why can't I edit this layer? It's a tile, imagery, map image, or KML layer — or editing is disabled on the item (Chapter 10)
Why no pop-ups? Tile, vector tile, and media layers carry no queryable attributes
Why did my shapefile land in the ocean? No .prj in the zip — Chapter 3
Cheapest way to map a spreadsheet? Add lat/lon columns before upload; geocoding addresses spends credits
Best upload format overall? Zipped file geodatabase
CSV came in with wrong field types? Fix types during the upload's field-review step, or re-publish — Chapter 11

Sharing and Permissions Checklist

Sharing in ArcGIS Online is per-item, not per-app. An app can be public while its map or layers are private — the app loads, the content doesn't. Work the chain from the bottom up. For hosted layer mechanics see Compendium Chapter 10 (Hosted Feature Layers); for admin roles and privileges see Compendium Chapter 34 (Administration).

The sharing chain

Every viewer must be able to reach every link. One private link breaks the whole chain.

Level What it is Where to check
App Instant App, Dashboard, StoryMap, Experience App's item page > Share
Web map / scene The map the app displays Map's item page > Share
Layers Each hosted or referenced layer in the map Each layer's item page > Share
Referenced services Layers served from Enterprise or external servers The source server's own security settings
Locator / utility services Custom geocoders, routing, print services used by the app Each utility item's Share setting

Rule: the sharing level of every item must be equal to or broader than the app's audience. Public app means public map means public layers.

Chain pre-flight:

Group sharing steps

For sharing with a specific team instead of the whole org or public.

Master + view pattern for public data

Never share your editable master layer publicly. Publish a read-only view instead. Details in Compendium Chapter 10 (Hosted Feature Layers).

Editable-layer exposure check

Run this on every layer shared beyond your team. Anonymous editing on a public layer means anyone on the internet can add, change, or delete your features.

Watch out: Sharing an editable master layer to Everyone is the single most damaging mistake in this topic. It hands the public full write access to your production data — vandalism, mass deletes, and junk features, often unnoticed for weeks. Always publish a read-only view for public consumption and keep the master private.

Public-launch checklist

Final gate before you send the link out.

Fast diagnostics

Symptom Likely cause
App loads, layers missing or erroring A layer shared narrower than the app
Login prompt on a "public" app Map or a layer still private, or subscriber/premium content in the map
Group member can't see the app Item not shared to the group, user isn't actually a member, or member is from another org and the item is shared only org-wide
Public user sees edit tools Editing enabled on the shared layer or view
Layer visible but fields missing View field restrictions — check whether that's intended

More failure patterns in Compendium Chapter 39 (Troubleshooting Encyclopedia).


Field Apps Chooser and Offline Checklist

Three mobile apps, three metaphors: Field Maps is map-first, Survey123 is form-first, QuickCapture is button-first. Pick by workflow, not by habit. Full treatment of all three: Compendium Chapter 30 (Field Maps, Survey123, QuickCapture). End-to-end walkthrough: Compendium Chapter 37 (Worked Project - Field Collection).

Decision matrix

Factor Field Maps Survey123 QuickCapture
Core metaphor Map with editable layers Standalone smart form Big buttons, one tap per record
Form complexity Moderate: conditional visibility, calculated expressions, required fields Highest: full branching logic, repeats, constraints, multi-page forms (XLSForm or web designer) Minimal: each button writes preset values, with at most a quick single-entry prompt
Map-centricity High. You navigate to and select features on a map Low. Map appears only as a location question Low. Location captured automatically in the background
Capture speed Moderate: tap map, fill form Slowest: work through the form Fastest: single tap, usable from a moving vehicle
Edit existing features Yes: inspect, update, move geometry Limited: existing records via the Inbox only No. Capture-only
Related records Yes: view and edit related tables from the parent feature Repeats publish as related records; editing existing ones is limited Not practical
Attachments Photos, video, and files from the device, per feature Photos, audio, signatures, sketches, image annotation Photos tied to the button press
Offline Yes: pre-planned or on-device map areas Yes: forms work offline; add an offline basemap for context Yes: projects run offline by default
Typical jobs Asset inspection, maintenance, inventory against known features Structured surveys, permits, inspections with heavy logic and signatures Windshield surveys, damage tallies, wildlife counts, incident pins

Ten-second chooser

Offline pre-flight checklist

Run this before any crew leaves the building. Web vs. Pro editing behavior: Compendium Chapter 13 (Editing Workflows).

Layers and map

Form and schema

People and devices

Watch out: Never change a layer's schema, disable sync, or overwrite the service while crews hold offline areas. Doing so invalidates their downloaded copies, and unsynced field edits can become unrecoverable. Freeze the schema for the life of the field campaign; if a change is unavoidable, have every device sync in first, then re-download after the change.

Top five sync gotchas

# Gotcha Symptom Prevention
1 Schema changed or sync disabled mid-campaign Sync fails outright; edits stranded on devices Freeze schema during fieldwork. Sync everyone in before any layer change (see Watch out above)
2 Basemap won't download Offline area stalls or errors before crews leave Use an export-enabled basemap or sideload a tile package; limit the download to the zoom levels the job actually needs
3 Attachment bloat Sync hangs or fails on cellular; huge upload queues Set photo size in the app, require Wi-Fi for photo-heavy syncs, split large offline areas per crew
4 Crews never sync Days of edits on one device; office map stale; device loss = data loss Enforce end-of-shift sync as policy; enable auto-sync where connectivity allows; monitor last-sync per user
5 Two crews edit the same feature Later sync silently wins; earlier edit disappears Assign non-overlapping work areas; use assignment fields or editor tracking; review edits daily while memories are fresh

Sync error messages and their fixes are catalogued in Compendium Chapter 39 (Troubleshooting Encyclopedia).

Quick reference: what each app cannot do

App Hard limits to remember
Field Maps Form logic is simpler than Survey123; no repeats inside the form itself (use related tables)
Survey123 Weak at browsing or editing the existing feature map; Inbox is the only path to existing records
QuickCapture Essentially form-free: preset values plus at most a quick prompt; no editing, no related records; wrong tool the moment users must answer real questions

Schema Design Checklist

Use this before you publish a layer or hand a form to field crews. Full treatment: Compendium Chapter 12 (Schema Design). Schema decisions are cheap before data exists and expensive after.

Field-type selection

You are storing Use Why / notes
Category from a known list Text + coded value domain Enforces valid values; renders as a dropdown in forms
Free-form notes Text, generous length Growing length later is usually possible; shrinking is not safe
Small whole numbers (counts, ratings) Short integer Smallest adequate type
Large whole numbers Long integer Population, IDs without leading zeros
Measurements, percentages, money Double Skip Float; the savings are not worth precision surprises
Dates and times Date Never store dates as text — you lose sorting, filtering, and time analysis (Compendium Chapter 20)
Yes/no Text or short integer + two-value coded domain Most geodatabase formats have no true boolean field
IDs with leading zeros (parcel, phone, ZIP) Text Numeric types strip leading zeros permanently
Unique record identifier GlobalID System-maintained; required for offline sync and used as the key for attachments and relationships
Photo, document Attachments (not a field) See attachments decision below

Watch out: A field's type is permanent once data exists. You cannot convert a populated text field to a date or number field in place — the only fix is add a new field, calculate values across, repair every map, form, pop-up, and app that referenced the old field, then delete it. Storing dates or numbers as text is the single most expensive schema mistake. Decide types before the first row exists.

Naming rules

Rule Detail
Treat field names as permanent Hosted feature layers offer no rename; a geodatabase rename tool exists but still breaks every map, form, and query that used the old name. Get the name right first — the alias is the safe thing to change
Lowercase, underscores, no spaces install_date, not Install Date!
Start with a letter Never a number or underscore
Keep names short Some formats truncate long names on export (shapefile is brutal — Compendium Chapter 11)
Avoid SQL reserved words date, order, user, desc, add cause query errors later
Set a human-readable alias on every field Aliases drive pop-ups, forms, and legends; they are cheap to change anytime
No PII in field names Names leak into URLs and REST endpoints (Compendium Chapter 32)

Coded domain vs free text

Choose coded domain when Choose free text when
The value list is known and finite (status, material, species) Genuinely unpredictable content (comments, descriptions)
You will filter, symbolize, or count by the value The value is never used for filtering or symbology
Multiple people will enter data One careful editor, throwaway dataset
Field crews use forms (dropdowns beat typing — Compendium Chapter 30) The list would exceed practical dropdown length

Rule of thumb: if you will ever chart or filter by it, use a domain. "Oak / oak / OAK / Oaktree" in free text costs a cleanup project later; adding a code to a domain costs a minute.

Defaults and required flags

Setting Use it for Caution
Default value The most common value (status = "Active", condition = "Unknown") A wrong default gets silently accepted at scale — prefer "Unknown" over a plausible-looking guess
Required (non-nullable) Fields the record is meaningless without Every required field slows field entry; require only what you would reject a record over
Domain + default together High-volume field collection Fastest possible form entry

Editor tracking and GlobalIDs

Attachments: yes or no

Enable attachments when photos or documents belong to a specific feature (pole photo, permit PDF). Skip them when:

Enabling attachments later is possible, but decide before designing field forms so crews capture photos from day one.

Signal Verdict
One value per feature, ever (install date, owner) Field on the feature
Many records per feature over time (inspections, readings, maintenance visits) Related table
You keep adding inspection1_date, inspection2_date... Stop — that is a related table
Data edited by a different team on a different schedule Related table

Relationships need a key: GlobalID on the parent, GUID field on the child. Design them together — retrofitting keys onto loose tables is manual work. Details in Compendium Chapter 12.

Cost of change after data exists

Change Cost
Field alias, pop-up, symbology Trivial — change anytime
Add a field, add a domain code Easy
Add a domain to an existing free-text field Moderate — must clean existing values first
Enable attachments, editor tracking Moderate — no backfill of history
Increase text length Usually possible
Rename a field, change its type, shrink text length Effectively a rebuild — not supported in place on hosted layers, and even where a geodatabase tool can rename, every dependent map, form, and app breaks. Plan on add/calculate/delete plus repairs
Change geometry type or the layer's spatial reference Republish from scratch (Compendium Chapter 10)

12-point pre-collection checklist

  1. [ ] Every field's type checked against the table above — no dates or numbers stored as text
  2. [ ] Field names lowercase, no spaces, no reserved words; final (not placeholders)
  3. [ ] Every field has a plain-language alias
  4. [ ] Every categorical field has a coded value domain, including an "Unknown" or "Other" code
  5. [ ] Defaults set on high-volume fields; no misleading defaults
  6. [ ] Required flags limited to genuinely mandatory fields
  7. [ ] GlobalIDs enabled
  8. [ ] Editor tracking enabled
  9. [ ] Attachments decision made and configured
  10. [ ] Repeating data moved to related tables with keys defined
  11. [ ] Test features created, edited in the real form (Compendium Chapter 13), then deleted
  12. [ ] Schema reviewed by one person who did not design it

If all twelve pass, publish. If any fail, fix them now — every row collected multiplies the cost of the fix.


Styling Decision Guide

Match your data's shape to a style family first, then pick classification and color. Full click paths and every style option: Compendium Chapter 7 (Styling and Smart Mapping). Color theory and visual hierarchy: Compendium Chapter 4 (Cartographic Design).

Step 1: What shape is your data?

Your data Style family Notes
Categories (land use, status, type) Types (unique symbols) Qualitative colors. Merge rare categories into "Other" if you have many.
Counts / totals (population, crash count) Counts and Amounts (size) Counts get size, not color. Colored polygons of raw counts mislead.
Rates / ratios (percent, per capita, density) Counts and Amounts (color) — choropleth Only for normalized values. Set the divisor in the style pane or precompute the field.
Value with a meaningful midpoint (change, +/- from average) Above and Below Diverging ramp centered on zero, average, or a target.
Dense point data (thousands of overlapping points) Clustering or binning (aggregation), or heat map See Step 4 table.
Totals inside polygons, shown as texture Dot density One dot = N units. Zoomed-out (small-scale) views only.
Two variables Color and Size, or Relationship (bivariate) Strongest combo: rate as color + count as size. Bivariate grids need a 10-second legend read — test on someone.
"Which category wins here" Predominance Compares competing count fields per polygon; opacity = strength of the win.
Location only, no attribute Single symbol (Location) Don't invent meaning; one symbol is often the right answer.

If a field can be styled with a computed value instead of raw data (a ratio, a category from thresholds), write an Arcade expression rather than adding a field — Compendium Chapter 8 (Arcade from Zero to Fluent).

Step 2: Classification method — each one has a bias

Method How it splits Its honest bias Reach for it when
Natural breaks (Jenks) Minimizes variance within classes Finds "groups" even in smooth data; breaks are dataset-specific, so two maps aren't comparable One standalone map, clumpy data, no better reason
Quantile Equal feature count per class Manufactures visual contrast; near-identical values land in different classes Rankings; evenly spread data; you want every class populated
Equal interval Equal-width value ranges Skewed data dumps most features into one class; map looks empty Familiar ranges (percentages, temperature); legend must be simple
Standard deviation Distance from the mean Assumes roughly normal data; legend baffles general readers Analyst audiences; "how unusual is this place"
Manual You choose breaks Your bias — which is fine when breaks come from policy or domain thresholds Regulatory cutoffs, round numbers, comparing maps across dates or regions
Unclassed (continuous) No breaks; smooth ramp Hides exact values; hard to cite a number Showing overall pattern, not lookup

Checklist before committing:

Step 3: Color ramp rules

Ramp Use for Rule
Sequential (light to dark, one hue) Ordered values, low to high Dark = more on light basemaps; flip when the basemap is dark. Never break this.
Diverging (two hues meeting at neutral) Values around a meaningful midpoint The midpoint must mean something (zero, average, target) — otherwise use sequential.
Qualitative (distinct hues) Unordered categories Keep it to about seven distinct hues; beyond that, group into "Other." No hue should look "higher" than another.

Colorblind rule: never let red vs. green carry the only meaning. Prefer ramps flagged colorblind-friendly in the ramp picker, and vary lightness as well as hue so the ramp survives grayscale. When in doubt, run the map through a color-vision simulator before publishing.

Watch out: The single most damaging styling mistake is a choropleth of raw counts — coloring polygons by totals when the polygons differ in size or population. Big empty counties look "high" simply because they're big. Normalize (per capita, per square mile, percent) or switch to size-based symbols. No classification method or ramp can rescue an unnormalized count map.

Step 4: Dense points — clustering vs. binning vs. heat map vs. dot density

Technique What the reader sees Best when Avoid when
Clustering Countable circles with numbers; expands on zoom Readers need to drill into individual features; interactive web maps Exact positions matter at every scale; static exports
Binning Fixed geometric cells colored/labeled by count Consistent area-to-area comparison; very large layers (server-side aggregation helps performance — Compendium Chapter 10) Bin edges would imply false boundaries in a small dataset
Heat map Smooth density surface First-glance "where is activity concentrated" at small scales Readers need values or counts; few points (a dozen points make a fake hotspot); print output. Weight by a field if magnitude matters, or it shows point density only
Dot density Scatter texture inside polygons Totals across regions, showing composition with multiple fields Zoomed in — dot placement is random within polygons and implies false precision

A heat map is a visual impression, not analysis. For statistically defensible hot spots, use the pattern-analysis tools in Compendium Chapter 17 (Statistical and Pattern Analysis).

Ten-second pre-publish check

Working in Pro instead of Map Viewer? Same decisions apply; the mechanics live in Compendium Chapter 23 (Symbology and Labeling).


Troubleshooting Triage Card

Run the three checks in order. Most failures die at check 1 or 2. If all three pass, go to the referenced chapter for the full diagnosis tree.

Before anything else: the 60-second sweep

The triage table

Symptom First three checks, in order Full diagnosis
Layer invisible (in the list but nothing draws) 1. Layer visibility toggle on, including any parent group layer.
2. Visible range — zoom in and out past the scale thresholds.
3. A filter or definition query is excluding every feature.
Compendium Chapter 39 (Troubleshooting Encyclopedia); styling causes in Chapter 7 (Styling and Smart Mapping)
CSV points in the wrong place (ocean off Africa, wrong country) 1. A pile of points at 0,0 — "Null Island," in the ocean off West Africa — means blank or unparseable coordinate values, not a projection problem.
2. Latitude and longitude columns mapped correctly at import — not swapped — and values in decimal degrees, not DMS text or projected coordinates; watch commas used as decimal separators.
3. Correct coordinate system chosen at import time.
Compendium Chapter 11 (Creating Data); coordinate theory in Chapter 3 (Coordinate Systems and Projections)
Cannot edit (tools greyed out, edits rejected) 1. Editing is enabled in the layer's item page settings.
2. Your account has edit privileges and access to the layer.
3. You are on the source layer or an editable view — not a read-only view or a copy.
Compendium Chapter 13 (Editing Workflows); views in Chapter 10 (Hosted Feature Layers)
Sharing broken (others get errors or a blank map) 1. Every layer in the map is shared at the same level as the map itself — update sharing from the map item, which flags mismatched layers.
2. The recipient is in the group or organization you shared to.
3. The map includes subscriber or premium content that demands sign-in from public viewers.
Compendium Chapter 10 (Hosted Feature Layers); admin side in Chapter 34 (Administration)
Slow map 1. Feature count and geometry complexity of the heaviest layer — simplify or generalize.
2. Visible ranges — detailed layers should only draw when zoomed in.
3. Optimization settings on the layer's item page (drawing optimization, caching/tiles for static data).
Compendium Chapter 10 (Hosted Feature Layers, performance section)
Misaligned data (consistent offset from basemap) 1. Datum mismatch — compare each dataset's geographic coordinate system.
2. Correct datum transformation applied in the map or during projection.
3. Define-vs-project confusion: was the wrong coordinate system assigned to data that actually needed reprojecting?
Compendium Chapter 3 (Coordinate Systems and Projections)
Sync conflict (offline edits missing or overwritten) 1. Read the sync error in Field Maps before assuming loss — retry the sync.
2. Did two editors change the same feature? The last edit synced wins by default.
3. Sync and offline settings on the layer item — were they changed mid-project?
Compendium Chapter 30 (Field Maps, Survey123, QuickCapture)
Geocode wrong (addresses land in the wrong spots) 1. Address fields mapped to the right locator fields — city or ZIP stuffed into the street column is the classic.
2. Country/region setting on the locator matches your data.
3. Match scores — review low-score candidates and rematch instead of accepting everything.
Compendium Chapter 11 (Creating Data, geocoding paths); credit cost in Chapter 2 (The ArcGIS Ecosystem)
Pop-up empty (or shows the wrong fields) 1. Pop-ups are enabled on that specific layer — not just the map.
2. Pop-up field list — fields hidden or removed in the pop-up configuration.
3. Arcade expression errors — test each expression in the editor.
Compendium Chapter 9 (Pop-ups, Fields, and Labels)
Credits drain (balance dropping faster than expected) 1. Org credit usage report — which tool, which user, which item.
2. Feature storage — bloated attachments, abandoned hosted layers, duplicate copies.
3. Repeated geocoding jobs or scheduled analysis running unattended.
Compendium Chapter 2 (The ArcGIS Ecosystem); budgets and allocation in Chapter 34 (Administration)
Analysis fails (tool errors out or hangs) 1. Inputs: valid geometries, overlapping extents, feature counts within reason for the tool.
2. Your account has the privilege and credits for that tool.
3. The actual error text in the tool results or geoprocessing history — not just the red X.
Compendium Chapter 39 (Troubleshooting Encyclopedia); per-tool detail in Chapters 16–20 (Vol D)
Login/license tangle ("no license," wrong content, can't sign in) 1. Right destination: your organization's URL, not the generic sign-in or a different org.
2. Your user type and licenses cover the app you are opening — check with your admin.
3. In Pro: open the Portals page in the project's backstage settings — confirm the correct portal is set active and you are signed in.
Compendium Chapter 2 (The ArcGIS Ecosystem); assignment fixes in Chapter 34 (Administration)

Watch out: Never "fix" a broken hosted feature layer by deleting it and republishing. The new item gets a new ID, which silently breaks every map, app, dashboard, form, and field deployment pointing at the old one — and can destroy field-collected data that only lived in that layer. Fix layers in place (overwrite, update definition, edit settings). See Compendium Chapter 10 (Hosted Feature Layers).

If all three checks pass: escalate with evidence

Collect these before filing a ticket or asking a colleague. It halves the round trips.

Full symptom-by-symptom trees live in Compendium Chapter 39 (Troubleshooting Encyclopedia).