Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Meta Ads › Campaign Structure

Bid Strategies on Meta

How Meta bid strategies work and how to choose one: highest volume, highest value, a cost per result goal, a ROAS goal or a bid cap, set from your economics.
Guideintermediate12 min readUpdated

What a bid strategy tells the delivery system

Meta runs an auction for each impression, and the pillar's account of delivery is the starting point: the winner is the ad with the highest total value, which combines the advertiser's bid, the estimated action rate and ad quality. The bid in that sum is rarely a figure anyone types. It is set directly only under a bid cap; under every other strategy it is implied, calculated by the delivery system from the budget, the optimisation goal and whatever goal the strategy carries. The Meta Ads hub describes the bid as what the result is worth to you, set directly or implied by the bid strategy, and that is the right way to hold it in mind.

The bid strategy therefore answers a narrow question: how hard should delivery push, and against what limit? It does not choose the result. That is the job of the objective and the optimisation goal beneath it, which the campaign objectives guide covers as the first decision in the campaign structure hub; a perfectly chosen bid strategy on an ad set optimised for link clicks will still find people who click. Nor does the strategy change how you are billed. The objectives guide explains that billing is mostly per impression whatever the ad set optimises for, with a per-action option for a small number of goals, so the cost per result (CPA) is the figure the strategy is really steering.

The strategies Meta offers

The site's Meta guides describe five strategies from Meta's documentation. The names in Ads Manager have shifted over the years and some options depend on the objective, so the interface is the authority on what a given ad set can select.

StrategyWhat it asks delivery to doWhat it needsThe trade
Highest volumeSpend the budget and get as many results as possibleNothing beyond a budget and an optimisation eventNo cost guarantee; the cost of results rises as spend grows
Highest valueSpend the budget and get as much conversion value as possible, in Sales campaignsPurchase values passed correctly on every purchase, and enough purchasesFewer results, chosen for value; meaningless if values are missing or wrong
Cost per result goalKeep the average cost of results near a figure you setA goal derived from the economics and a history to judge it againstSpends less, or not at all, when results cost more than the goal
Return on ad spend goalKeep return on ad spend (ROAS) above a minimum, for value optimisationReliable purchase values and a return goal derived from marginSpends less when the auction cannot deliver the return
Bid capNever bid above a set amount in any auctionA firm view of what an impression is worth, and time to adjust itStrict control of the auction price, frequent adjustment, little control of cost per result

Two things in the table deserve emphasis. The constrained strategies, the bottom three, share a failure mode: a goal or cap below what the auction requires does not produce cheap results, it produces very little delivery. The objectives guide puts it plainly, and the troubleshooting page for ads not spending budget shows how it looks in an account: results that come in under the goal while spend sits far below budget. And the bid cap is different in kind. The scaling guide notes that it controls the price paid in each auction rather than the cost per result, which suits advertisers who know what an impression is worth to them and want strict control, at the price of frequent adjustment.

Start on highest volume, and why

A new ad set should almost always start on highest volume. The reason is the learning phase. Delivery needs a flow of optimisation events to learn whom to show the ad to, and Meta's documentation describes the period after an ad set is created or significantly edited as unstable, with cost per result usually higher than it will be afterwards. A constraint applied at that moment cuts the flow of events the ad set needs to settle. The learning phase guide lists a cost goal or bid cap on a new ad set among its common mistakes, and lists bid or cost controls set too tight among the causes of the learning limited status. The objectives guide makes the same point: constrained bidding with no history starves delivery.

There is a second reason, which is information. A goal is a statement about what a result costs in the auction this ad set faces, and nobody knows that before the ad set has run. Highest volume reveals it. Run the ad set on the default, leave efficiency judgements until it has settled, and read the cost of results over a full week or longer. That figure is not the goal. It is the evidence against which a goal, derived from the economics, can be checked for realism.

Highest volume is not a reckless choice for an account with a firm cost ceiling. The budget is the limit on spend, and an ad set whose cost of results is over target on highest volume has a problem a bid strategy will not fix: the creative, the offer, the event or the audience. A cost goal on that ad set would simply stop it spending, which hides the problem rather than solving it.

Setting a cost or return goal from the economics

When the business does need a guardrail, set it from what a result is worth, never from the current average. The scaling guide is explicit about why: delivery buys the cheapest results first, so the next result always costs more than the average one, and a cap copied from last month's average chokes delivery as soon as spend grows. The ads-not-spending page traces most strangled ad sets to exactly that origin.

For a cost per result goal, start from the most the business can pay for one result and still meet its margin. For a purchase, that is the contribution an order makes after cost of goods, fulfilment and payment fees, less the profit the business wants to keep. For a lead, it is the value of a lead: the share of leads that become customers, multiplied by what a customer is worth, less the same profit allowance. The CPA calculator does the arithmetic once those inputs are known.

For a return on ad spend goal, start from Break-even ROAS, which the break-even ROAS entry defines as one divided by the gross margin: the return at which ad spend exactly consumes the margin. The goal sits above it by the profit the business wants after advertising. The ROAS calculator turns spend and revenue into the ratio, and the same reasoning in reverse turns a margin into the floor.

Then check the goal against the evidence from highest volume, using marginal rather than average cost. The scaling guide sets out how: compare two periods at different spend, and the extra spend divided by the extra results is what the last results cost. If the economic limit sits comfortably above that marginal cost, a goal will mostly act as insurance. If it sits below, the goal will cut delivery hard, and the business is learning that this ad set cannot buy results at a price it can afford, which is worth knowing before raising the budget.

Count or value: when highest value and ROAS goals fit

Highest volume counts results; highest value weighs them. The objectives guide describes highest value as the strategy for Sales campaigns that maximises the value of conversions when purchase values are passed, and the Meta Ads setup for online sales sets out when it is worth using: when order values vary widely, when bundles or premium lines pull the average order up, when the value parameter is correct on every purchase and when there are enough purchases for the ad set to learn. A shop whose orders are all a similar size gains little, because every result is worth about the same.

The two questions to settle before switching are data and volume. The value must be right: a purchase event that sometimes carries no value, or carries a value in the wrong currency, teaches delivery to chase the wrong people, and the sales setup says plainly that value optimisation on a broken value parameter optimises for nonsense. And value optimisation needs purchases to learn from, so an ad set that struggles to leave learning on purchase count will not do better on purchase value. The sales setup's advice is to start on count, check the values, and test value optimisation in its own ad set.

A return on ad spend goal is the value-side guardrail, the equivalent of a cost goal for count. It fits when value optimisation already works and the business needs a floor under return, for instance before raising the budget. The comparison of maximise conversions and target CPA bidding describes the same trade between spending the budget and holding a cost on Google Ads; the logic carries across, the mechanics do not, so read each platform's own documentation for how its strategies behave.

Bid strategies under campaign budget

With budgets on each ad set, each ad set can carry its own strategy and goal. With Advantage+ campaign budget the strategy moves up a level. The campaign versus ad set budgets comparison, which draws on Meta's documentation of campaign budget, sets out the consequences: every ad set beneath the campaign shares the campaign's bid strategy, and cost and return goals apply across the campaign rather than to each ad set. Under highest volume the system moves spend towards whichever ad set delivers results most cheaply; under a cost per result goal it balances spend across ad sets to hold the campaign's average near the goal; under value strategies the same logic applies to value.

Three practical points follow. First, a campaign goal is judged on the campaign's average, so an individual ad set can run above it while another runs below; read each ad set's cost of results, not only the total. Second, the ad sets under one campaign budget need to be genuinely comparable, optimised for the same event and producing results the business values equally, because one strategy and one goal will govern all of them. Third, the budget types guide notes that switching an existing campaign to campaign budget is subject to conditions such as the ad sets using compatible bid strategies, which is one more reason to settle the strategy before the structure rather than after.

Changing strategy without losing a working ad set

Meta's learning phase documentation treats a change of bid strategy, bid cap or cost goal as a significant edit, and the learning phase guide includes it in its table of edits that put an ad set back into learning. An ad set that has settled at an acceptable cost is an asset, and changing its strategy in place spends that asset for an experiment.

The lower-risk pattern is the one the learning phase guide recommends for any significant change: duplicate the ad set, apply the new strategy to the copy, let the copy settle, then compare and pause whichever loses. Give the copy a budget of its own so that it is not starved before it has learned, and run both long enough to read results on the metric that matters, which for a value strategy is revenue and return, not the count of purchases. The cost is a period in which two ad sets compete for similar people; the benefit is that the account never goes without a stable ad set.

When an in-place change is unavoidable, make it once and batch it with any other significant edits, so the ad set learns once rather than several times. Choose the strategy before a scaling push rather than during it; the scaling guide's checklist asks that the bid strategy is chosen before a budget step and not changed with it, because two edits produce one result and no way to tell which caused it. And if an ad set with a new goal stops spending, the ads-not-spending page is the place to start: a goal below the going rate is the most common reason, and the fix is to raise it in one change or return to highest volume.

A yarn shop moves from volume to a ROAS goal (illustrative)

Illustrative example — not a real result

Thimblecrow Yarns moves from volume to a return goal

Thimblecrow Yarns, an invented yarn and knitting-kit shop, sells everything from single skeins at £9 to full kits at £120. Its gross margin after goods, packing and payment fees is 50%, so break-even return on ad spend is 1 ÷ 0.50 = 2.0. The owner wants to keep 12% of revenue as profit after advertising, so ad spend may be at most 38% of revenue, and the target return is 1 ÷ 0.38, about 2.6.

Period and set-upSpendPurchasesCost per purchaseAverage purchase valueRevenueReturn
Four weeks, highest volume, £50 a day£1,40070£20.00£44£3,0802.2
Next four weeks, original ad set, highest volume£1,40072£19.44£43£3,0962.2
Same four weeks, duplicate ad set, highest value£1,40049£28.57£79£3,8712.8
Following four weeks, highest value with a return goal of 2.6, £100 a day£2,60084£30.95£82£6,8902.65

Highest volume was above break-even but below target, and its buyers were mostly single-skein customers. The duplicate on highest value bought fewer, dearer purchases and returned more, so the owner paused the original. With a month of value history, she added a return goal of 2.6 taken from the margin, not from the duplicate's 2.8, and doubled the budget. The goal held spend £200 below budget over the four weeks, which the owner accepted as the price of keeping return above target.

Common mistakes

  • Starting a new ad set on a cost goal or bid cap. It cuts the delivery the ad set needs to settle; start on highest volume.
  • Copying a goal from the current average. The next result costs more than the average one, so the goal chokes delivery as soon as spend grows.
  • Using highest value with unreliable purchase values. Delivery learns to chase whatever the broken values reward.
  • Expecting a goal to fix an ad set that is too expensive. A constraint stops spend; it does not repair the creative, offer or event.
  • Changing strategy in place on a working ad set. It restarts learning; test the change in a duplicate.
  • Reading a campaign-level goal as an ad set guarantee. Under campaign budget the goal applies to the campaign average.
  • Changing strategy and budget together. Two edits, one result, no way to tell which mattered.

Frequently asked questions

What is the difference between a cost per result goal and a bid cap on Meta?

A cost per result goal asks delivery to keep the average cost of results near a figure you set, letting individual results cost more or less. A bid cap limits what delivery may bid in each auction, which controls the price of an impression rather than the cost of a result. Both are constraints, and both stop delivery when set below what the auction requires; the cost goal is usually the easier one to derive from a business's economics, because it is expressed in the same unit as the target.

Should a new ad set start with a cost cap?

Usually not. A new ad set has no cost history for the goal to be judged against, and the site's learning phase guide lists a cost goal or bid cap on a new ad set among the common mistakes, because it suppresses the delivery that would have produced the events the ad set learns from. Start on highest volume, read the cost of results once the ad set has settled, and add a goal derived from the economics if the business needs a ceiling.

Does changing the bid strategy reset the learning phase?

Yes. Meta's learning phase documentation treats a change of bid strategy, bid cap or cost goal as a significant edit that can put the ad set back into learning. On an ad set that is working, the safer route is to duplicate it, apply the new strategy to the copy, let the copy settle and then pause the original, so the account always has one stable ad set running.

Terms used on this page

Tools to try

Sources

  1. Meta Business Help Centre — Meta Business Help Centre
  2. About the learning phase — Meta Business Help Centre
  3. About Advantage+ campaign budget — Meta Business Help Centre

Related