Where bids and budgets are set
Three levels matter. The insertion order carries the budget, split into dated segments if the plan needs them, and its own pacing mode and rate. It can also carry optimisation settings that move budget between its line items or apply a shared bid strategy across them towards the insertion order's performance goal; the Help Centre describes the conditions under which that insertion-order-level automation is available. The line item carries its own budget, either an amount within the insertion order or unlimited up to what the insertion order allows, its own pacing, and the bid strategy that is actually used in the auction. The campaign carries a planned spend used for pacing comparisons and nothing that bids.
Partner costs and fees are added on top of the media price and do not change what DV360 bids, but they do change what a budget buys. A line item whose insertion order carries several partner costs delivers fewer impressions for the same budget, and a constrained automated strategy sees that higher delivered cost as its own. The account structure guide explains where each of these settings lives and how they are applied downwards; this guide takes the structure as given and covers how the settings behave once live. For the auction underneath, how a bid is formed and cleared in real-time bidding (RTBAuction per impression in milliseconds.), the real-time bidding guide is the reference.
Fixed vs automated bidding
| Aspect | Fixed bid | Automated bidding |
|---|---|---|
| What you set | A maximum CPMCost per 1,000 impressions. for every impression, optionally adjusted by bid multipliers per targeting value | A goal to maximise, an optional target or ceiling, and a budget to spend |
| What DV360 decides | Nothing beyond whether the request matches targeting | The bid for each request, from predicted value |
| Signals used | None | Floodlight conversions, clicks, viewability or completed views, depending on the goal |
| Best for | Deals at an agreed price, early learning, inventory with stable prices | Performance goals at scale, line items with steady conversion volume |
| Risk | Overpaying in cheap auctions and losing expensive ones | Learning periods, drift when signals are thin, hidden trade-offs between goal and budget |
| Transparency | Full | Partial; bid-level decisions are not exposed |
| Changes | Immediate | Each change restarts learning |
Fixed bids are the right default for deal line items and for the first flight of any new inventory source, because they expose what the inventory clears at. Automated strategies exist in DV360 in two broad families: maximise a goal while spending the full budget, with an optional ceiling on the average cost, and meet or beat a target for a cost metric such as cost per acquisition (CPACost of each conversion.), where the target takes priority over spending the budget. The Help Centre lists the current goals for each line item type. Custom bidding, where an advertiser supplies its own scoring logic built on Floodlight and impression signals, is a third option for teams with the data to feed it.
Goals, constraints and the learning period
Every automated strategy is a goal plus a constraint, and the two pull against each other. "Maximise conversions while spending the full budget" treats spend as the fixed point and lets cost per conversion float; adding a ceiling on average cost per thousand impressions keeps the bids within a band at the price of some volume. "Beat a target cost per action" treats the target as fixed and lets spend float, which is why a line item on that setting can under-deliver for weeks with no error shown anywhere: it is doing what it was told. Choose the setting from what the plan actually promised, volume or efficiency, and read under-delivery on a target strategy as the strategy's verdict on the target.
After any change to the strategy, its target or the line item's conversion events, DV360 needs time to re-learn, and the Help Centre advises against judging a strategy or changing it again during that period without committing to a fixed length. In practice the signal is stability: when the daily cost per goal event stops swinging, the learning has settled. Learning needs volume, so thin line items with a handful of conversions a week rarely settle at all; consolidating them into fewer, larger line items is often the fix. The Smart Bidding guide describes the same discipline in Google Ads, and the habits transfer even though the strategy names do not.
Pacing modes and daily vs flight budgets
Pacing has two parts in DV360: a mode and a rate. The mode is daily or flight. Daily pacing takes the budget as an amount per day; flight pacing takes the budget as a total across the segment's dates and works out the daily allowance itself. The rate is as soon as possible, even or ahead. As soon as possible spends whenever matching requests arrive until the period's allowance is gone; even spreads the allowance across the period; ahead front-loads so that the line item runs slightly ahead of an even schedule, which the Help Centre describes as a hedge against late-flight shortfalls.
Both the insertion order and each line item carry these settings, and both apply. An insertion order on even flight pacing throttles the combined spend of its line items to its own daily allowance regardless of what the line items ask for, which is how a deal line item can be starved by an open-auction sibling that reaches the shared allowance first. Giving the deal its own insertion order, or setting line item budgets as explicit shares rather than unlimited, removes that contention. Daily pacing on a line item inside a flight-paced insertion order is a common and workable combination; the reverse rarely is.
Budget segments on the insertion order let one order carry a different budget for each date range, so a plan that changes allocation month by month does not need a new insertion order each time. The segment's end date is also an end date for everything under it, which is the first thing to check when spend stops on the first of the month. The CPM calculator is the quickest way to turn a segment's budget into the impression volume a fixed bid implies, and from there into whether the targeting can supply it.
Frequency caps
Frequency (FrequencyAverage times a person saw an ad.) caps limit how many times one person is served over a time unit, and DV360 lets them be set at campaign, insertion order and line item level, each as a number of impressions per period from minutes up to the life of the object, or as no limit. Every level's cap is enforced for the impressions under it, so caps stack: the campaign's cap bounds the whole plan, the insertion order's cap bounds its channel or budget line, and the line item's cap bounds the tactic. The usual pattern is a loose cap at campaign level to protect the person from the plan as a whole, and tighter caps on remarketing line items. The DV360 campaign launch checklist confirms caps, pacing and bid strategy against the plan before the flight begins. Choosing the cap level itself is a question the ad frequency cap test answers with capped and uncapped cells read on lift per unit of spend.
Caps depend on recognising the same person across impressions, and that recognition is partial where identifiers are missing. The Help Centre describes how DV360 manages frequency in environments without cookies or device identifiers, including modelled approaches, and those details change often enough that the help page is the source rather than a summary here. For how caps and reach relate across platforms, and why a DV360 cap does not bound what the same person sees elsewhere, the reach and frequency across platforms guide takes up the measurement side. If a cap is set and people still see the ad far more often than it allows, the diagnosis of a frequency cap that does not hold separates line-item-only caps from the exposures no DSP cap can see.
Diagnosing under-delivery
The line item troubleshooting view reports, for a date range, the bid requests the line item was eligible for, the bids it submitted, the auctions it won and the impressions it served, alongside any flags DV360 raises about the line item itself. Read the four figures in order; the first one that collapses relative to the one before it names the problem area. When the collapse points at creatives, structure, a deal, an audience list or an inclusion list rather than the bid, the diagnosis of a DV360 line item that is not spending routes each of those causes to the guide that owns it.
| Where the funnel narrows | Common cause | Check |
|---|---|---|
| Few bid requests | Targeting too narrow, inventory source not active, brand-safety or viewability filters removing most inventory, deal not sending | Each targeting dimension removed one at a time; deal status |
| Requests but few bids | Line item or insertion order paused or outside dates, no approved creative in a matching size, budget or pacing allowance already consumed, frequency cap reached for most matched users | Status and dates, creative approval, pacing view, cap settings |
| Bids but few wins | Fixed bid below clearing prices; automated strategy constrained by a ceiling or target that the inventory cannot meet | Raise the fixed bid or loosen the constraint; compare against the win rate of a sibling line item |
| Wins but few impressions | Creative rejected by an exchange after the auction, or a rendering failure | Creative status per exchange |
| Spend on plan but goal events low | Strategy learning, conversion events not firing, or audience and inventory that simply do not convert | Floodlight test, learning stability, segment-level reports |
| Spend front-loaded then stopped | As soon as possible pacing, or an insertion order allowance consumed by one line item | Pacing rate and line item budget shares |
Turning around an under-delivering app-install line item
A fintech advertiser runs a DV360 display line item with a £40,000 four-week flight budget on even pacing, an automated strategy set to beat a £9 target cost per sign-up, and a cap of 3 impressions per person per day. After ten days it has spent £6,200. The troubleshooting view shows 48 million eligible bid requests, 4.1 million bids and 210,000 wins. Requests are plentiful, so targeting is not the issue; the collapse is between bids and wins, and the sign-up Floodlight activity is recording only 14 conversions a day, far too few for the strategy to learn from. The trader switches to maximise conversions while spending the full budget with a £6 average CPM ceiling, raises the cap to 5 per day, and consolidates two sibling line items that shared the same audience into this one. Wins rise to 1.3 million over the next week, sign-ups climb to 60 a day, and the line item finishes the flight at £38,900 spent with a cost per sign-up of £8.40.
When the plan's aim is to be remembered rather than to convert, the DV360 brand awareness setup applies these bid strategies, caps and pacing modes to reach and viewability goals.
To make the daily pacing read a routine with agreed limits on who may move budget, the programmatic campaign management playbook sets it beside weekly inventory and frequency reviews and a monthly reconciliation.
Common mistakes
- Changing the strategy and the target in the same week. Learning restarts and nothing can be attributed.
- Reading under-delivery on a target strategy as a bug. The strategy is declining to pay above the target; the decision is whether the target was right.
- Unlimited line item budgets under one insertion order. The hungriest line item takes the allowance and the deal line items starve.
- Forgetting fees when setting a ceiling. The ceiling bounds media cost; the budget pays media plus partner costs plus data fees.
- Caps only on the line item. Each line item serves up to its own cap, and the person sees all of them.
- Judging a fixed bid without the win rate. A low win rate with plentiful requests says the bid is low; the same impressions figure with few requests says the targeting is narrow.
- Setting as soon as possible pacing on a flight budget. The first days consume what the month needed.
The DV360 hub places bidding among the other line item decisions, and the inventory and creatives guides there cover the two settings that most often turn out to be the real cause of a bid problem.
