8 min read
Cross-Channel Bid Optimization: Where Should the Next Advertising Dollar Go?
Use marginal return, comparable business outcomes and realistic tests to decide where more Google Ads budget can help.
Cross-channel bid optimization starts with the return you expect from the next dollar, not the average return on the dollars already spent. A campaign with excellent historical ROAS may have little room to grow. Another may look weaker on average but have worthwhile opportunities left.
Compare useful business outcomes, understand what each report counts, and distinguish a forecast from a measured result. Then decide whether to add budget, move it from somewhere else, change a bid target, or leave the account alone. Those are different decisions.
High average ROAS does not tell you what more spend will buy
Imagine two campaigns over a comparable, mature reporting period. Search spent $10,000 and recorded $60,000 of purchase revenue: a 6× ROAS. Demand Gen spent $20,000 and recorded $80,000: a 4× ROAS. It is tempting to move money from Demand Gen to Search.
But the decision concerns a change in spend. Suppose a forecast suggests an extra $2,000 in Search would produce $3,000 more revenue. Its forecast marginal ROAS is 1.5×, not 6×. Suppose removing $2,000 from Demand Gen is expected to lose $5,000 in revenue. If those forecasts were accurate and comparable, the move would reduce total revenue by $2,000 at unchanged total spend.
These are invented numbers, not Google benchmarks. Their point is simple: a good average can coexist with an unattractive next opportunity. The same reasoning applies to leads, provided you compare a useful lead outcome rather than treating every form submission as equally valuable.
| Hypothetical change | Spend change | Revenue change |
|---|---|---|
| Add to Search | +$2,000 | +$3,000 |
| Cut Demand Gen | −$2,000 | −$5,000 |
| Combined forecast | $0 | −$2,000 |
Make the numbers comparable enough to use
Before accepting that calculation, ask what “revenue” means in each row. Is it purchase value or an assigned value for a lead? Are cancellations and returns included? Do both reports cover the same currency and period? A neat spreadsheet does not fix incompatible inputs.
Attribution also matters. A person can discover a product in one place and buy after interacting with another. Reports assign credit according to their settings; they do not automatically reveal how many sales would disappear if you removed a campaign. Nor should you add independently attributed platform totals and call the sum unique customers.
Demand Gen's Conversions (Platform Comparable) columns offer a separate reporting view. They do not change bidding or make different platforms' attribution methods identical. Use them for the question they answer, with their windows and interaction rules visible—not as interchangeable replacements for the usual Conversions column. The platform-comparable guide explains the distinction.
Conversion delay can make a recent campaign look worse simply because more of its customers have not bought yet. Compare periods with enough time for the relevant outcomes, or clearly label the recent numbers as incomplete. Use the account's observed delay; video does not automatically deserve an arbitrary extra month, and Search is not always immediate.
When measurement is uncertain, show the range of plausible outcomes rather than forcing a precise ranking. A budget decision can still be made under uncertainty, but it should not pretend the uncertainty has disappeared.
Find out what is actually limiting the campaign
More budget helps only if the campaign can use it productively. A campaign that regularly spends below its budget may instead be constrained by its target, available demand, targeting, creative or product availability. Raising the budget alone does not create more good opportunities.
For eligible Search campaigns, lost impression share from budget can help identify missed reach. It is not a forecast of profitable sales. Lost impression share from rank is a different problem, and neither metric means that all lost impressions deserve to be bought.
Performance Max adds another distinction: its campaign budget covers multiple inventories. A campaign-level budget increase does not reserve the extra money for Search. Examine the channel report when that distribution matters, but do not treat it as a set of independently editable channel budgets.
Google's Performance Planner models eligible Search, standard Shopping, Performance Max, Demand Gen and App campaigns. Display and Video planning ended March 9, 2026. Check the campaign-specific requirements before using a forecast: Demand Gen, for example, excludes product-feed campaigns. The Demand Gen table requires seven days without a bid-strategy change, while the page's generic exclusion says ten; use the actual eligibility result rather than promising access on day eight.
A scenario is an estimate, not a commitment from Google. Review the proposed settings before applying a plan—application can change live campaigns. Keep a forecast and the later observed result distinct; a planning tool's precision does not make its assumptions certain.
Check the receiving side and the side you plan to cut. Moving money away from a promotion that has just started, or into products about to run out, may be a bad business decision even if last month's campaign ratios favor it. Stock, sales capacity and contribution margin belong in the decision because the ad report may not contain them.
Test the decision you are actually making
Separate the hypothesis from the action. “Search has higher ROAS” is an observation. “Another $2,000 in this Search campaign will earn more contribution than we lose by cutting that Demand Gen campaign” is a testable budget hypothesis.
Choose a scale the business can tolerate and enough observation time for the relevant conversions and returns. There is no universal 15% budget step or seven-day test that makes every reallocation safe. A small account and a mature high-volume account will produce different amounts of useful evidence.
Where an eligible experiment or a well-designed market test can isolate the change, prefer that to reading two arbitrary months side by side. If a before-and-after comparison is the practical option, record major differences such as promotions, inventory, pricing and measurement changes. They may explain some or all of the result.
Review the combined business outcome as well as campaign reports. In the hypothetical above, Search could report growth while the business loses revenue overall. Conversely, one campaign's attributed ROAS might fall while total profitable orders rise. Neither pattern proves causality on its own, but both deserve investigation.
Finally, let the evidence change the decision. More budget is not the only successful outcome. You might keep the allocation, adjust the target, repair the landing experience or wait for inventory. The useful result is a better choice, not a budget edit made to demonstrate activity.
Questions and answers
Should I always increase the campaign with the highest ROAS?
No. Its historical average does not show the return on more spend. Consider the additional outcome it could generate, the outcome lost by funding the increase elsewhere, and the uncertainty in those estimates.
Is moving budget the same as changing a bid target?
No. Budget determines the spending allowance; a CPA or ROAS target guides the bidding objective. Understand which constraint is limiting the campaign before deciding what to change.