Why online sales changes the DV360 setup
A sales plan is the one DV360 plan whose outcome has a price. An order carries a value and a quantity, and Campaign Manager 360 (CM360Ad server and measurement platform.) records them only on a sales activity; a counter activity, the type used for forms and sign-ups, records that something happened and nothing more. Track purchases with a counter and the account can report a cost per conversion but never revenue, and standard counting adds every reload of the confirmation page.
Attribution is the second difference. Display is seen far more often than it is clicked, and the order activity credits DV360 with purchases that follow an unclicked impression inside its view-through window, so that window decides how many orders display claims. The third is that a sales plan buys two different groups of people: strangers, and shoppers who viewed a product and left. The second group buys far more readily, so a line item that mixes them flatters prospecting and hides what remarketing adds.
The online sales objective hub rates programmatic as a supporting route and judges every channel on contribution after ad cost, on return on ad spend compared with the break-even figure margin sets, and on the share of orders from new customers. The setup below gives those figures honest inputs from DV360: orders, their value and the customer status recorded on each. A plan that has to make the brand known rather than sell needs a different build: the DV360 setup for brand awareness puts exposure goals, campaign-level caps and even pacing where the order activity sits here.
Recommended DV360 setup for online sales
| Setting | Recommended for online sales | Why it differs from other objectives | Feature page |
|---|---|---|---|
| Conversion activity | A sales group for orders: one activity on the confirmation, transactions counting, order number in ord, value in cost, quantity in qty | Only a sales activity carries the order value; lead and awareness plans count or have no event at all | Floodlight tags |
| Windows and counting | A short view-through window set on purpose; post-click orders as the measure, post-view on its own line | Unclicked impressions can claim orders that would have happened anyway | DV360 reporting |
| Line items | One insertion order per budget line; prospecting, remarketing and each deal on separate line items | Remarketing buys people already near an order, so its cost per order must not blend with prospecting | Account structure |
| Audiences | Product-viewer and basket lists created when the tags go live; recent purchasers excluded from prospecting | Lists collect only from creation, and a sales plan pays again for buyers it does not exclude | Audiences in DV360 |
| Creatives | Dynamic creatives from the product feed on remarketing, keyed on the viewed product, with a default row | Only a sales plan has a catalogue to show back to the shopper | Creatives in DV360 |
| Bidding | Fixed CPM on deals and new inventory; then maximise conversions on the order activity under a CPM ceiling; a target cost per order last | The order is the goal event, and automated strategies need a steady flow of it to learn | Bidding and pacing |
| Inventory | Open-exchange prospecting on an inclusion list; deals with publishers whose readers buy the category | Prospecting buys strangers in the open auction, where exclusions alone admit sites nobody chose | Building an inclusion list |
Record the order as a sales activity
The Floodlight tags guide covers the choice every later number depends on.
Create a sales-type group for orders, since a group's type cannot change later, with one activity on
transactions counting: the order number in ord, so a refreshed thank-you page stays one order, the value in
cost and the item count in qty, all read from the data layer. Forms and sign-ups stay in counter groups, out
of the revenue columns. Spend custom variables on customer status and product category, which later reports and
lists use. The tutorial on setting up a new advertiser links
DV360 to the Campaign Manager 360 advertiser that holds this configuration and records a test order before launch.
Read post-click orders first and post-view orders on their own line
Each activity carries its own click-through and view-through windows, and the Floodlight guide names a long view-through window on cheap display as the biggest single source of inflated counts. Keep the order activity's view-through window short and record it as a decision. The DV360 reporting guide splits Floodlight conversions into post-click and post-view: judge line items on the first and report the second beneath it. The comparison of view-through and click-through conversions recommends an incrementality test before giving view-through conversionsA conversion credited to an ad that was shown but not clicked, within a set view window. a value. If the Floodlight count still runs above the orders in the business's records once the window is short, the diagnosis of Floodlight conversions counted twice separates repeated firings, a tag fired twice, shared credit and a doubled goal.
Split line items by tactic
The account structure guide makes an insertion order one budget line and a line item one tactic. A sales plan needs at least three: open-exchange prospecting on Google audiences, remarketing on Floodlight lists and one line item per deal, each with its own cost per order, bid and cap. Give deals their own insertion order or an explicit budget share, as the bidding and pacing guide advises, so the open exchange cannot spend the shared allowance first, and put prospecting on an Inclusion ListA list naming the sites and apps a programmatic buy may run on, excluding everything else by default; sometimes called an allowlist. built as the inclusion list tutorial describes.
Build product lists early and exclude buyers
A Floodlight list collects members only from the day it is created, as the DV360 audiences guide explains, so create the product-viewer, basket and purchaser lists when the tags go live, with membership durations that match how long people take to choose. Exclude recent purchasers, and Customer Match uploads of existing customers where eligible, from every prospecting line item through one combined audience, and give the remarketing line item a tighter frequency cap than prospecting.
Show the viewed product on remarketing
The creatives guide describes dynamic creatives rendered from a feed, built in Ad Canvas or in Studio with a dynamic profile. Key the remarketing feed on the product the person viewed, which the Floodlight tag carries in a custom variable, set a default row for impressions no rule matches, and treat the first flight as a test of what rendered. Prospecting keeps static creatives in every size the inventory sends.
Move bidding from fixed prices to the order
New inventory and deals start on fixed bids, which show what they clear at. Once the order activity records conversions on a line item at a steady rate, move it to the strategy that maximises conversions while spending the budget, under a ceiling on the average cost per thousand impressions. A target cost per order (CPACost of each conversion.) comes last: a target strategy holds the target and lets spend float, so a target below what the inventory delivers leaves the line item under-delivering with no error shown. The CPA calculator splits a cost per order into CPM, click-through rate and conversion rate, which shows which lever moved when a line item's cost per order changes. Merge line items that see only a handful of orders a week; custom bidding, scored from Floodlight and impression signals, is the third option the bidding guide lists, suited to advertisers who hold the data it needs.
A worked online sales setup
Plumwexa Luggage, an invented online retailer of suitcases and travel bags
Plumwexa Luggage sells suitcases and travel bags online at an average order of £118; its margin puts break-even ROAS at 2.5. It spends £28,000 a month in DV360. The first build, one open-exchange line item bidding on a counter activity, reported 1,310 conversions at £21.37 each, but about one firing in six was a reload of the thank-you page and 930 of the conversions were post-view.
| Setting | First build | Rebuild |
|---|---|---|
| Conversion activity | Counter on the confirmation page, standard counting | Sales group "Orders": transactions counting, order number in ord, customer status in a custom variable |
| Windows | Left as created; post-view and post-click read together | 30 days click-through, 1 day view-through; post-view on its own line |
| Line items | One open-exchange line item | Prospecting, remarketing and one preferred deal |
| Audiences | In-market travel audience and site visitors together; no exclusions | Product-viewer and basket lists on remarketing; purchasers of the last 90 days excluded from prospecting |
| Creatives | Four static sizes | Dynamic product creative on remarketing; six static sizes on prospecting |
| Bidding | Maximise conversions on the counter | Fixed £6 CPM on the deal; fixed £4 CPM on prospecting and remarketing in the first month, then maximise conversions on orders with a £5 CPM ceiling from the second month |
| Inventory | Open exchange, advertiser defaults | Prospecting on an inclusion list of 380 sites |
Two months on, the same budget produced 404 post-click orders at £69 each: prospecting spent £17,000 for 168 (£101 each), remarketing £8,000 for 205 (£39 each) and the deal £3,000 for 31 (£97 each). Sales-activity revenue was £47,700, a ROAS of 1.7 overall and 3.0 on remarketing, with 310 post-view orders on their own line. First-time buyers placed 72% of prospecting orders against 18% on remarketing, so the team cut prospecting to £13,000 and booked a geographic holdout to measure what it adds.
The rebuild did not improve the plan's numbers; it made them readable. The old figure mixed duplicate firings, unclicked impressions and returning shoppers, and the bid strategy had been learning from all three.
Measuring online sales in DV360
Start with DV360's own columns. Post-click conversions and post-view conversions on the order activity are the core pair, read under the lookback windowsThe time after an ad interaction in which a conversion can still be credited to it. the Floodlight configuration sets. Cost per conversion depends on the cost metric: media cost is what the exchanges charged, partner costs and data fees sit on their own lines, and the reporting guide leaves the make-up of total media cost to the Help Centre and asks for one cost metric per rate, written down. The sales activity's value fills the revenue and return columns; DV360's own revenue metric is what the advertiser is billed. The Floodlight report in Campaign Manager 360 gives the raw activity count against which DV360's attributed share is read, and the diagnosis of Floodlight not recording conversions starts there when either falls. List sizes and the remarketing line item's frequency show whether remarketing still has people to reach.
Beside them sit the sales hub's measures: return on ad spend (ROASRevenue generated per unit of ad spend.) by tactic against break-even ROASMinimum ROAS to avoid a loss., which the ROAS calculator derives from margin; New-Customer ShareThe proportion of conversions or orders that come from first-time customers, flagged from the business's own records rather than a platform's report. for whether prospecting widens the customer base; and contribution after ad cost and holdout revenue for whether DV360 belongs in the plan at all. The measurement plan framework records which cost metric, window and report each figure comes from, and RACE applied to online sales sets these measures beside the other stages. Check tag health and pacing weekly, cost per order and ROAS by line item every fortnight, list sizes and placements monthly, and run a holdout each quarter.
Where the sales setup goes wrong
The attributed count covers only DV360's own media, so it cannot be added to other platforms' claims on the same orders, and consent and browser limits shrink both the lists and the matched orders. Remarketing always looks cheapest because it reaches people already close to buying; without a holdout its low cost per order overstates what it adds. Changing the order activity and the bid strategy in the same week restarts learning and leaves nothing attributable. And a dynamic creative whose feed has no default row renders blank or is rejected for the impressions where no rule matches.
Common mistakes
- Tracking orders with a counter tag. No revenue is recorded, and standard counting adds every reload.
- A random value in
ordon the sales tag. Each refresh of the confirmation page becomes another order. - Post-view and post-click orders added together. Display takes credit for purchases it may not have caused.
- Prospecting and remarketing in one line item. The blended cost per order hides which tactic pays.
- Lists created on launch day. Remarketing has nobody to reach for weeks.
- A target cost per order on a new line item. With no history the strategy under-delivers.
- Reading DV360's revenue metric as sales. It is the billed cost, not the value of the orders.
