> ## Content Index > Fetch the complete content index at: https://cascader.io/blog/llms.txt > Use this file to discover other available public pages before exploring further. # Your account shrank after switching to profit bidding. Should you switch back? - URL: https://cascader.io/blog/profit-bidding-account-shrank-switch-back/ - Published: 2026-11-08T15:00:00.000Z - Updated: 2026-11-08T15:00:00.000Z - Description: A smaller account after moving to profit bidding isn't proof the move failed. Four checks, worked through on a constructed two-product store, show what shrank it and whether switching back would fix anything. - Author: Rick Hecker - Tags: Cascader Answers, Google Ads, Smart Bidding, Conversion value Last verified October 2026 You changed the conversion value in your value-based bidding from revenue to profit, and the account got smaller. Spend is down, the bidder favors different products, and orders and new customers have fallen. Yet ROAS looks fine, and the campaigns aren't losing money. The natural conclusion is that profit bidding doesn't suit your account and you should switch back. That conclusion skips a step. First, be sure which ROAS figure looks fine: after the switch, the ROAS column measures whatever value you now send. Check 1 below tests the profit figure itself. Once it passes, three causes within profit bidding can produce these symptoms, and checks 2, 3 and 4 test them in turn: - a target that became stricter when the value changed, because the number in the box didn't move with the value (check 2); - profit bidding doing its job: it stops buying orders that cost more in advertising than they contribute (check 3); - profit bidding missing a number: a first order's profit leaves out what a new customer buys later (check 4). Each calls for a different response. The first needs the target re-priced, and the third needs the missing value added. The second needs no fix, only an honest scoreboard. A budget change or broken data can shrink an account too, and the checks below route those as well. We recommend going back to revenue only if check 1 shows the profit figure can't be made sound. Run the checks in order, because each one assumes the one before it passed. Throughout, "profit" means contribution: what an order leaves after the cost of goods, discounts, shipping you pay, payment fees and returns, before advertising. Some guides call bidding on it POAS, profit on ad spend. The worked example is a constructed store, with numbers invented to make the arithmetic visible. ## What actually changes when the value switches to profit Swapping revenue for contribution as the conversion value bundles two changes. The first is a rescale. Every purchase value is multiplied by roughly your average margin: at a 30% margin, $100 of revenue reports as about $30\. By itself that changes nothing about which auctions the bidder wants, as long as the target is multiplied by the same factor. Google describes Target ROAS as aiming for conversion value per cost that averages out at the target, with individual conversions landing above it and below it (Google: About Target ROAS bidding). A fixed add-on, such as a customer acquisition value, isn't multiplied at all; check 4 deals with it. The second is a reweight. A product with a below-average margin now carries less value relative to the rest of the catalog, and an above-average one carries more. That is the point of profit bidding, and it will cost some orders on purpose. The constructed store shows both. It sells two products, and in one matured month before the switch it took: - 20 orders of a starter set at $50\. Each contributes $7.50, a 15% margin, and 60% of these buyers are new customers. - 10 orders of a premium set at $100\. Each contributes $45, a 45% margin, and 20% of these buyers are new. That month brought $2,000 of revenue and $600 of contribution, a 30% margin across the mix. The campaign bid on revenue at a 500% Target ROAS. ## Check 1: can your profit figure carry bids? Every later check uses the profit figure, so test it first. Look at four things. **It includes every cost that differs by order.** That means discounts, cost of goods, shipping you pay, payment fees and returns. A figure without returns or shipping overstates exactly the products that get returned or are costly to ship, and the bidder will chase them. Test it by setting one matured month of profit values in Google Ads against the same orders' contribution in your books. **It is stable.** If costs are updated so often that the same product reports a different value week to week, the bidder is chasing movement in your spreadsheet rather than in your customers. Savvy Revenue's guide lists heavy promotions as a reason to skip or delay profit bidding, because "large discounts push per-transaction profit close to zero." **There is enough of it.** For Target ROAS on Search and Shopping, Google requires at least 15 conversions in the past 30 days, counted at the conversion tracking level (Google: About Target ROAS bidding). **Margins actually differ.** If margins are nearly flat, profit values are revenue multiplied by a near-constant, and the reweight has nothing to work with. In that case the mix shouldn't have moved at all. If it did, the cause is in check 2 or in the data. If the figure fails one of the first three tests for a reason you can fix, such as returns or shipping left out, fix it and run check 1 again before deciding anything else. Go back to revenue as the bidding value only if the figure can't be made sound. Keep reporting profit alongside, as a Secondary action, and base your revenue targets on the margins it shows. Products with very different economics can get their own campaigns and targets; [our guide to Target CPA and Target ROAS](https://cascader.io/blog/google-ads-target-cpa-target-roas/) works through that split. If you repair the figure later, run check 1 again before switching back to profit. ## Check 2: did the target get stricter when the value changed? This test is arithmetic. Take the target in force since the switch and divide it by the pre-switch ratio of contribution to revenue for the campaign whose target you're testing; for a portfolio strategy, use the ratio across all the campaigns it governs. If the result is higher than that campaign's or portfolio's old revenue target, the target got stricter. For "the target in force", use the Avg. target ROAS column over the dates you're judging. Google defines it as the traffic-weighted average of the targets that applied over that period (Google: About Target ROAS bidding). For the store, the equivalent profit target is 500% × 30% = 150%. You can get the ratio in two ways: - If you reported profit as a Secondary conversion action before the switch, divide its value by the revenue value for that campaign or portfolio over the same pre-switch dates. Savvy Revenue's guide notes that custom columns must use "All conversion value" to pick up a Secondary action. - Otherwise, take the same orders' revenue and contribution from your store's records over the same pre-switch dates, where they show which campaign produced each order. If you have neither, work the ratio across all your orders from those dates and treat it as an approximation: it is exact when the campaign's revenue-weighted margin matches the account's, as it does when the campaign sells the same product mix, and the further the campaign's margins sit from the account's, the less you can trust it. Use the mix from before the switch: afterward the mix has moved, and so has the ratio. If a customer acquisition value, or any other fixed-value Primary action, was active before and after the switch, the switch didn't multiply it by the margin, and Google's Conversion value column includes it alongside the purchase value (Google: Configure your lifecycle goals). Work the ratio on purchase value without it; your store's records give you that directly, where they show which campaign produced each order. For a customer acquisition or win-back value, Google also reports the amount it added in the "New Customer lifetime value" and "Win-Back customer lifetime value" columns, so you can take that amount out of the conversion value; if it appears in a Secondary profit action too, take it out there as well. Then don't re-price the target yet: the add-on needs resizing too, and "Add the new-customer value in the same unit" in check 4 shows how, and what that does to the target. Now see what each possible target asks for. The table gives the target-meeting cost of one order for each product: the advertising cost per order at which that product's orders exactly meet the target, which is its value per order divided by the target. | Value sent and target | Starter order | Premium order | Spend the same 30 orders allow | | ------------------------------------------------------------- | ------------- | ------------- | ------------------------------ | | Revenue, 500% (before the switch) | $10.00 | $20.00 | $400 | | Contribution, 150% (equivalent) | $5.00 | $30.00 | $400 | | Contribution, 300% (= 1,000% on revenue) | $2.50 | $15.00 | $200 | | Contribution, 500% left unchanged (= about 1,667% on revenue) | $1.50 | $9.00 | $120 | Smart Bidding doesn't price orders one at a time like this. It holds an average, so read the table for direction, not for bids. At the equivalent target, the same month supports the same $400\. It moves from the starter set, which can now cost only half as much, to the premium set, which can cost half as much again. At either stricter target, the money available falls. Don't let your product reports overrule the sum. It is tempting to look at your high-margin products and read a healthy reweight if they held up. But a moderately stricter target lets them hold up too. At a 200% contribution target, a third stricter than the equivalent and 667% on revenue: - the premium order may cost $22.50, up from $20; - the starter order falls to $3.75; - spend on the same orders falls to $300. The premium order keeps at least its old $20 allowance at any contribution target up to 225% (750% on revenue). So high-margin products holding up fits both a healthy reweight and a tightening. When the product picture and the arithmetic disagree, the arithmetic governs. A picture that contradicts it sends you to check the budget and the data, not to overrule the sum. Neither can a revenue ROAS that held up while spend fell decide it, since a stricter target produces that as readily as a reweight does. If you switched gradually, expect the arithmetic to say "stricter" for a while. Savvy Revenue explains why it steps the target down rather than jumping: on day one, most of what the bidder has learned is still in revenue, so going straight to the profit target would push it to bid far harder than you meant. Its schedule lowers the target week by week, and it says that after roughly 30 days the revenue history has dropped out of what the bidder works from. ProfitMetrics' guide also lowers the target in stages. A step above the equivalent is stricter for as long as it lasts, and that is intended. The trap is a step-down that has stalled, or one still above the equivalent after the revenue history has aged out (about 30 days in Savvy Revenue's account). If you're past that point, or switched in one move, and the arithmetic says stricter, lower the target to the equivalent, or set it from the business's floor: the lowest ROAS it can accept on the value you report. [Our guide to Target CPA and Target ROAS](https://cascader.io/blog/google-ads-target-cpa-target-roas/) works out that floor in step 2 of "How to choose your starting Target CPA or ROAS", and its steps 5 and 6 choose the target against it. If the gap is large, move in steps, as both guides do, reading daily spend against its pre-switch level. Change nothing else meanwhile. Google says the bidder reacts to a new target immediately but needs one to two conversion cycles to reach it (Google: About Target ROAS bidding), so judge after that, on matured data. Don't go back to revenue at this stage, because profit bidding hasn't had a fair test yet. Two cases need a note here: - **No target (Maximize conversion value).** There is no number to convert. Google's strategy table presents this one as the pick when you want the whole budget spent to get as much conversion value as possible. A rescale can't tighten it, so if spend fell, check for budget changes and go back over check 1\. The reweight in check 3 still applies. - **"Limited by budget".** This status doesn't take a target campaign outside the test. Google says budget-limited campaigns on a target-based strategy "now perform more consistently toward your bid target, including when you make budget adjustments" (Google: Changes to target based bid strategies). Run the arithmetic above as for any target campaign. ## Check 3: were the orders you lost making money? If the arithmetic says the target is equivalent, or you've re-priced it and waited, the bidder is pulling back from low-margin products by design. Before you call that a loss, price what those orders used to cost. Under revenue bidding at 500%, a starter order met the target at $10 of advertising. It contributes $7.50\. Every starter order bought at that price left the store $2.50 worse off before overhead. Across the constructed month: - starter orders: 20 × −$2.50 = −$50; - premium orders: 10 × ($45 − $20) = +$250; - kept after $400 of advertising: $200. Half the month's spend went on starter orders that lost money, carried by the premium set's surplus. Revenue bidding with one target can't see that. Profit bidding can, and it stops paying for it. When that is what happened, lower spend, lower revenue and more contribution kept can all be true at once. The lost revenue was revenue you were paying to have. To check your own account, take the products that lost the most spend. For a matured window before the switch, compare what their orders cost in advertising with what those orders contributed. One limit applies before you start. Google's standard product reports attribute each sale to the clicked product, which may not be the item the customer purchased. Savvy Revenue tracked a single product for 14 days and found that 55% of the resulting buyers purchased a different item from the one they had clicked. If you send conversions with cart data, Google can report the purchase itself: "Standard reports show you that a purchase was made, but cart conversion reports tell you exactly what the customer bought." That works only for items sold from a Merchant Center account connected to your Google Ads account (Google: About conversions with cart data). So pair the advertising cost of a product's clicks with the contribution of the orders those clicks produced, whatever those orders contained, taking that contribution from cart data where you have it. Without cart data, and unless your own records tie each order to the click that produced it, you can't build this pairing per product. Then treat the comparison below as direction only: a product that looks loss-making is a candidate for check 4, not a verdict. With the pairing in hand: - **Cost per order above contribution per order:** the bidder is dropping orders that lost money on the first purchase. Go to check 4 before writing them off. - **Cost per order below contribution per order, but above contribution divided by your profit target:** the orders made money but didn't meet the target. That's your target deciding. Leave it; or, if the business wants those orders, lower the target toward the floor-with-margin end of the band in our Target guide's step 5, which that step reserves for a business that wants the volume, a campaign not limited by budget and room to spend more; or lower the business's floor (the one check 2 names) on purpose, write down who agreed and why, and then set the target from the new floor by the guide's steps 5 and 6. - **Cost per order below both, and the product lost spend anyway:** with a target in force and no budget limit, the reweight doesn't explain it. Re-run check 2's arithmetic on the Avg. target for the window, then check the budget and the data. Where there is no target, or the budget is the binding limit, dropping orders that made money, or met the target, can be the reweight at work. The question there is whether to fund them with more budget. ## Check 4: what do the new customers from those orders bring back later? A first order's contribution leaves out everything a new customer buys afterward. If your low-margin products are where first-time buyers tend to start, profit values tell the bidder to cut the very orders that bring them in. Put one starter order on a ledger. Suppose each new customer's later orders contribute $20, on average, over the horizon you care about: - first-order contribution: $7.50; - advertising cost at the old price: $10.00; - kept today: −$2.50; - expected later contribution: 60% new × $20 = $12.00; - overall: +$9.50. Dropping that order saves $2.50 now and gives up $12 later. For any product the bidder is pulling back from, the rule is: > Keep paying for the order when the share of its buyers who are new, times the later contribution per new customer, is more than its cost per order minus its first-order contribution. For the starter set, the break-even is $2.50 ÷ 60% = about $4.17 of later contribution per new customer. Later contribution decides the rule, so build it carefully and err low: - **Source:** group customers by the month of their first order, and add up the contribution of their repeat orders at 90, 180 or 365 days, whichever horizon the business is willing to wait for its money. - **Deductions:** subtract the advertising that won those repeat orders, such as remarketing or brand clicks. - **Overlap:** the overlap that matters is customers who would have bought from you anyway. The ad didn't acquire them, so neither their first order nor their later ones belong in the later contribution per new customer. A holdout test is one way to measure that. A practical narrowing, which is our judgment rather than a Google rule, is to build the cohort only from first orders that came through non-brand campaigns, where your records show which campaign produced each first order. - **Default:** when you can't separate these, use the shortest horizon and the lower figure. An understated later contribution errs safe. It keeps fewer loss-making orders, and it adds a smaller acquisition value, which loosens the target less. How you identify a new customer matters as much as what one is worth; [our answer on what counts as a new customer in Google Ads](https://cascader.io/blog/what-counts-new-customer-google-ads/) covers the definition and Google's detection limits. If the rule says the orders were worth keeping, switching back to revenue fixes the wrong thing: revenue overpays for every low-margin order, not just the ones that bring customers back. Keep profit values and give the bidder the missing value. ### Add the new-customer value in the same unit as everything else Google's New Customer Value mode, part of the customer acquisition goal, adds an extra value, the customer acquisition value, to a new customer's first purchase conversion. Google requires the following (Google: About customer lifecycle goals): - **Bidding:** a value-based strategy (Target ROAS or Maximize conversion value). - **Goal:** at least one Purchase conversion goal. - **Campaigns:** it is available for Search, Shopping, Performance Max and Demand Gen. If your own systems know whether a buyer is new, you can instead add the value yourself, in the value you send. Do one or the other: doing both counts the new customer twice. Google's guidance on how big that value should be is written in revenue. It recommends setting the acquisition value so that it, plus a new customer's typical first purchase, "represents the revenue that a typical new customer would bring over their lifetime" (Google Ads Help: working together with AI). Its Accelerate with Google best practices suggest 2× average order value as a starting point when you're unsure; for the store that is 2 × ($2,000 ÷ 30) = $133.33\. In a profit-valued account, both need translating, and the table shows why. Each row gives the average value per order the bidder sees, the month's total value, and the spend that total allows at the 150% target. | Value sent | Starter order | Premium order | Premium ÷ starter | Month's total value | Spend allowed at 150% | | --------------------------------------------------------------------------------------- | ------------- | ------------- | ----------------- | ------------------- | --------------------------- | | Revenue | $50.00 | $100.00 | 2.0 | $2,000 | $400 at its own 500% target | | Contribution | $7.50 | $45.00 | 6.0 | $600 | $400 | | Contribution + $20 for each new customer | $19.50 | $49.00 | 2.5 | $880 | $586.67 | | Contribution + $133.33 for each new customer (2× average order value, a revenue figure) | $87.50 | $71.67 | 0.82 | $2,466.67 | $1,644.44 | For the new-customer rows, per-order values are averages: the starter set's $19.50 is $7.50 plus 60% of $20. **Contribution alone.** The premium set is worth six starter sets to the bidder. **The $20 later contribution, in the same unit.** It pulls that back to 2.5, still further toward the premium set than revenue did. **The revenue-sized value.** It flips the comparison. The starter set is now worth more than the premium set, and the month carries over four times the value it did on contribution alone. In the same unit as your profit values, the acquisition value should be a typical new customer's later contribution: $20 here, not $133.33\. If an acquisition value set in revenue terms was already active before the switch, your account is in that last row: replace it with later contribution in profit units, and make the target decision below in the same change as check 2's re-pricing. **The target still needs a decision.** Even the correct $20 raises the month's total value from $600 to $880, so the same 150% target now allows $586.67 of spend on the same orders instead of $400: 47% more. Adding new-customer value at an unchanged target is a loosening. Google's page permits basing your ROAS targets on that combined value once you use a value-based strategy (Google Ads Help: working together with AI). That makes the target a business decision, and the store shows two ways to make it: - **Spend up to two-thirds of the new total value and keep a third.** It keeps 150% and accepts up to $586.67 of spend on these orders. - **Hold spend at $400 while it learns whether new customers really return $20.** It sets $880 ÷ $400 = 220%. At 220%, a starter order meets the target at $8.86 and a premium order at $22.27; those are averages again, not bids. Either choice can be defended. Leaving 150% in place without choosing it can't. One limit applies. A single acquisition value treats every new customer alike. If buyers who start on one product come back far more than buyers who start on another, the setting can't say so. If that matters, add the value yourself, per product. ## What to do with what you found | What you found | What it means | What to do | | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | The profit figure is incomplete, unstable (including heavy promotions) or too thin | It can't carry bids yet | If the cause can be fixed, fix it and re-run check 1 before deciding. If it can't, go back to revenue as the bidding value, keep profit as a Secondary action, and set targets from margins by product group; re-run check 1 before switching to profit again | | The target in force divided by that campaign's (or portfolio's) pre-switch margin ratio is above its old revenue target, whatever your high-margin products did | The target is stricter | Inside a planned step-down, keep stepping. Otherwise lower the target to the equivalent or set it from the business's floor (see check 2), moving in steps if the gap is large. If an acquisition value is active, resize it first (check 4) and set the target on the new total. Change nothing else, and judge after one to two conversion cycles on matured data | | No target (Maximize conversion value), and spend fell | The target isn't the cause | Check for budget changes and re-check the profit figure, then go on to checks 3 and 4 for the mix | | The lost orders cost more than they contributed, and their new customers fail the rule in check 4 | Profit bidding is dropping orders that lost money | Keep profit bidding, and judge and report results as contribution minus advertising (next section). The value you send stays as it is | | The lost orders lost money on the first purchase, but pass the rule | The value is missing later customer worth | Keep profit values, add a new-customer value in profit units, and set the target on the new total | | The lost orders made money but didn't meet the target | Your target is deciding | Leave it; lower the target within the guide's band; or lower the business's floor on purpose and re-set the target from it (conditions in check 3) | | The lost orders met the target and lost spend anyway | With a target in force and no budget limit, the reweight doesn't explain it | Re-run the target arithmetic on Avg. target ROAS, then check the budget and the data. Where there is no target or the budget is the binding limit, decide whether to fund those orders with more budget | | Margins are nearly flat, and the mix still moved | Margins aren't the cause | Re-check the target arithmetic and the data | ## How to judge it afterward Wait long enough first. Google's page on changing conversion goals counts a change in value scale among the changes after which Smart Bidding takes "1-2 conversion cycles in most cases" to learn. It adds that Performance Max may take longer if most of its traffic comes from outside Search or Shopping (Google: Changing conversion goals and actions used for Smart Bidding). Its Target ROAS page gives the same one to two cycles for Search, framed as a wait before adopting a target on changed values. Compare windows that start after that. Don't judge the result on revenue, or on ROAS in either unit. Judge it on what the business keeps: contribution minus advertising cost for a matured window. Compare it with a window of the same length before the switch, in a similar season where you can, and set beside each the number of new customers it acquired, times their later contribution. In the constructed month, revenue bidding kept $200\. If profit bidding spends less and keeps more, even after you subtract the later contribution of any new customers it stopped acquiring, it is working, however much smaller the account looks. Before you change anything, write down: - which row of the table you're in; - the figure that put you there; - the date you'll read the result. The account that shrank after the switch may simply have stopped paying for orders it couldn't afford. Those notes are how you'll know.