Floodlight in SA360
Floodlight is a tag placed on the advertiser's site that records an activity (a page view, a form submission, a purchase) against a Floodlight configuration belonging to a CM360 advertiser. Activities come in two kinds: counter activities record that something happened, and sales activities also record a quantity and a revenue value. Custom variables can be passed with either kind to carry extra detail such as product category, order margin or lead type.
SA360 uses Floodlight by linking a client account to the CM360 advertiser that owns the configuration. From then on every activity in the configuration is available in SA360 as a conversion column and as a conversion source for bid strategies. The join between a conversion and a search click works through the click identifier that each engine appends to the landing-page URL and that the Floodlight tag reads when it fires. If the identifier is missing, because auto-tagging is off, a redirect strips parameters, or a tracking template does not produce the engine's parameter, the tag still records the activity for CM360 but SA360 cannot credit it to a click.
Three settings in the Floodlight configuration govern how conversions are counted in SA360. The lookback window sets how long after a click a conversion still counts, and because search is click-based the click-through window is the one that matters. The AttributionAssigning conversion credit to touchpoints. model decides how a conversion that followed clicks on more than one engine is credited; the default credits the last SA360-tracked click, and the help centre lists the models available. And the counting method on each activity (every instance, once per user per day, once per session, and the options specific to sales activities) decides whether repeat form submissions inflate the count.
Two documented behaviours are easy to forget. Floodlight conversions appear in SA360 after a delay, so the most recent days are always incomplete when a report is run and a bid strategy's recent window is always short. And a change to the configuration (a new activity, a changed window, a different model) applies from the change forward, so a report that spans the change compares two definitions. The current experience and the previous version differ in where some of these settings live and in column naming, so the help centre, not memory of the old product, is the reference for the current screens.
Conversion columns and custom metrics
| Column or metric | What it counts | Built from | Use it for |
|---|---|---|---|
| Floodlight activity columns | Conversions for one activity, joined to SA360-tracked clicks | The linked configuration | The base figure for any Floodlight KPI |
| Floodlight revenue and quantity | Value and quantity from sales activities | Sales activities | Value-based KPIs and ROAS |
| Engine conversion columns | Conversions as the engine itself counts them | Each engine's conversion tracking | Comparison with Floodlight; accounts without a CM360 link |
| Conversion goal columns | The conversions and value a bid strategy optimises towards | The goal's chosen sources and weights | Judging bid strategies on their own basis |
| Custom conversion columns | Floodlight conversions filtered by activity and custom variable values | Activities plus custom variables | Category revenue, lead quality tiers, margin |
| Custom columns | A formula over other columns | Any available metrics | Cost per qualified lead, revenue per click, blended KPIs |
| Engine delivery columns | Impressions, clicks and cost as the engine reports them | Sync from each engine | The denominators for every efficiency metric |
The distinction that matters most is between Floodlight columns and engine conversion columns. They count different things under different rules and will not agree; both are useful, and a report should say which one each KPIA metric tied directly to an objective. uses. The previous version grouped Floodlight data into standard actions, transactions and revenue columns; the current experience exposes activities as columns you add and combine, so the names on the current screens are the ones to use in documentation.
Custom conversion columns are where Floodlight earns its keep. A sales activity that passes product category in a custom variable can be split into one revenue column per category without any change to the tag; a lead activity that passes a quality score can be filtered to qualified leads only. Those columns can then feed a custom column (cost divided by qualified leads) and a bid strategy's conversion goal, so the figure the team reports and the figure the strategy optimises are the same. The bid strategies guide covers the goal side.
Custom columns follow the same logic as in Google Ads: a formula over existing metrics, saved once and available in every report. The discipline is to define each KPI once, name it unambiguously, and resist building near-duplicates that differ in one filter; a report with three columns called conversions is a reconciliation problem waiting to happen.
Cross-engine reports
A cross-engine report is any report at client account level or above whose rows span engine accounts. The report editor supplies the dimensions (engine, engine account, campaign, ad group, keyword, label, device, date) and the columns above, and the result is a single table in which a Google Ads campaign and its Microsoft Advertising mirror sit as two rows with the same measures. Segmenting by engine turns a total into a comparison; grouping by label turns mirrored campaigns into one logical row.
Three things make such a report trustworthy. First, labels or names that are consistent across engines, which is the point of the discipline in the cross-engine management guide: without them the report cannot group what belongs together. Second, a single conversion basis, Floodlight columns for every row or engine columns for every row and never a mix, because a mixed basis makes the engines look different for reasons that have nothing to do with performance. Third, a date range that ends before the Floodlight delay, so that the last days are not under-reported relative to the rest.
Reports can be saved with their filters and columns and re-run, and SA360's reporting extends upwards: a sub-manager or manager account report rolls client accounts into one view, which is how an agency sees every advertiser's search in one table and how a multi-market advertiser sees its markets. Time zones deserve a check at that level: engine accounts can be in different time zones and the help centre describes which time zone a report uses, so a day boundary in one market is not a day boundary in another.
The data also leaves SA360. The help centre documents a reporting interface for programmatic access and connectors for business intelligence tools, which is how a Floodlight figure reaches a warehouse or a dashboard that also carries display and video; the Looker Studio section covers that end. Anything exported carries the same caveats as the report it came from: the basis, the delay and the time zone travel with the numbers.
Executive summaries and scheduling
A summary report is a cross-engine report reduced to the figures a non-practitioner needs: spend, conversions or revenue on one agreed basis, the resulting CPA or ROAS, and a comparison with the previous period and the plan. In the current experience it is built from the same report editor and dashboards as any other report. The previous version had a separate executive report feature; whether a dedicated equivalent exists in a given account is a question for the current help centre, and the content is the same whichever tool builds it.
Scheduling is what turns a report into a process. A saved report can be scheduled to run and be delivered on a cadence, and a dashboard can be shared, so that the Monday figures arrive without anyone opening SA360. Two rules make scheduled reports safe. Schedule them to run after the Floodlight delay has passed for the period they cover, so the delivered figures are complete; a report that runs at midnight on the last day of the month will understate the month. And include the basis in the report itself, a note that conversions are Floodlight, last click, with the configured window, so that a reader who compares it with the engine's own interface knows why the numbers differ before they ask.
The summary is where the measurement plan does its work. A plan that names each KPI, its source column, its attribution basis and its owner, written before the first report circulates, is what stops the quarterly argument about which number is right; the summary then simply reports what the plan defined. Where a KPI's definition changes (a new activity, a changed window) the plan is updated and the summary notes the break in the series.
Reconciling with engines and analytics
SA360, each engine and the analytics tool will report different conversion counts for the same campaigns, and the differences have causes that can be named.
| Cause | Direction of the gap | How to check |
|---|---|---|
| Attribution across engines | SA360 credits the last SA360-tracked click; each engine credits its own click, so the engines together over-count relative to SA360 | Sum the engines' conversion columns and compare with the Floodlight total |
| Lookback windows | A longer engine window counts late conversions SA360's window drops, or the reverse | Compare the engine's window with the Floodlight configuration's |
| Click identifier missing | Tag fires without an identifier SA360 can join; CM360 sees the conversion, SA360 does not | Compare the activity's total in CM360 with its SA360 column |
| Modelled and cross-device conversions | The engine may include conversions it has modelled or matched across devices that the Floodlight tag never saw | Read the engine's conversion column definition |
| Tag placement and firing | Floodlight tag on a different page or firing condition from the engine tag | Test both tags on the conversion path |
| Counting method | One tag counts every submission, the other one per session | Compare the activity's counting setting with the engine action's |
| Date conventions and time zones | Conversions dated to the click or to the event; different account time zones | Read each column's definition in the help centre; align the report time zone |
| Consent and blocking | Tags blocked or declined at different rates on different paths | Compare tag firing rates per page in the tag manager |
| Analytics counting rules | Sessions not clicks; its own channel attribution; event-date reporting | The platform and analytics discrepancies guide |
The method is the same as for any pair of systems. Pick a period that is complete on every side. Rule out tag faults first, because a tag that is not firing is the one cause that is an error rather than a definition. Then attribute the remaining gap cause by cause, from the largest down, and record the expected size of the gap once it is explained. The Google Ads and GA4 comparison walks through that pair in detail, and the same logic applies between SA360 and GA4 with one extra step for the cross-engine attribution.
Once the gap is explained it becomes a monitoring metric: a gap that stays inside its usual range is a definition, a gap that moves is a finding. The commonest finding is a click identifier that has stopped arriving on one engine, after a tracking template edit or a new redirect, and it shows up as a Floodlight column falling while the engine's own column holds steady. When the Floodlight column reads zero outright, the diagnosis of Floodlight not recording conversions works from the ad server's own count through the link, the windows and the identifier to find which layer lost it.
Reconciling one month across SA360, two engines and GA4
For one month a travel company's non-brand search reports 1,240 conversions in Google Ads, 310 in Microsoft Advertising, 1,380 in the SA360 Floodlight booking column and 1,150 bookings attributed to paid search in GA4. The team starts with the engines: 1,550 engine conversions against 1,380 in SA360, a gap of 170. Testing shows the Microsoft tracking template was edited mid-month and dropped the click identifier for nine days, which accounts for about 90 conversions that CM360 recorded but SA360 could not join. Of the remainder, 50 are Google Ads cross-device and modelled conversions the Floodlight tag never saw, and the final 30 are bookings credited by both engines to their own click, which SA360 credits once to the last click. The GA4 figure is lower again because 230 bookings followed a journey whose last session came from email or organic search, which GA4 credits elsewhere. The template is fixed, the expected gap is recorded as roughly 5% against the engines and 15% against GA4, and the monthly summary states its basis.
Common mistakes
- Mixing Floodlight and engine columns in one KPI. The two count differently; a report that uses one for Google and the other for Microsoft makes the engines look different for no real reason.
- Reporting before the delay has passed. The latest days are always incomplete; a schedule that runs too early understates every period.
- Assuming the engine's count is the true one. Engine, SA360 and analytics each answer a different question; the measurement plan decides which answers which decision.
- Ignoring the click identifier. A tracking template edit that drops the parameter breaks the join silently; the Floodlight column falls while the engine's holds.
- Changing the configuration without noting the break. A new window or model from mid-month makes the series non-comparable, and nobody told the readers.
- Columns that nobody can define. Three columns called conversions with different filters, and a reconciliation that cannot start.
- Executive summaries without a stated basis. The reader compares them with the engine interface, finds a gap, and the credibility of the whole report goes with it.
The SA360 hub places measurement alongside management and bidding; the bid strategies guide is where the conversion columns defined here become the goals strategies optimise towards.
