Best Practices for Optimizing Facebook Bidding and Budgets

Patrik Akrenius

Are you always adjusting your bids and budgets, and having trouble finding the right levels while maintaining stable ad delivery? Fortunately, bids and budgets are coupled on Facebook through pacing, and we’ve mapped out what you should take into account to ensure optimal performance and have a setup that's easy to control.

This blog post covers the basics of budget pacing, lowest cost vs. target cost bidding, bid vs. budget limited campaigns, and how these work together with Smartly.io’s AI optimization tools.

How pacing works in practice

Prices in Facebook auctions vary due to competition on the auction platform. To avoid reacting too fast to price changes and wasting the whole budget suddenly, Facebook controls ad spend for smoother delivery through pacing. Pacing happens both when allocating a budget over multiple days and when distributing bids within a day. 

When you’re using a lifetime budget, budget pacing allows you to allocate a campaign budget evenly over multiple days. The goal is to avoid over- or under-delivery, and to use the budget completely and steadily over the time period.

Bid_Pacing_2018-01

Bid pacing controls both how daily budgets are used inside each day and how Facebook reacts to changing price levels in ad auctions. There are two options for bidding: lowest cost and target cost. Lowest cost bidding strives to get the lowest price in each auction and uses discount pacing for this. In discount pacing, all auctions are participated in, but bid levels are lowered to get the least expensive results available while striving to spend the budget. The bid cap is the maximum amount paid for any single conversion in the target audience.

Bid_Pacing_2018-02

Target cost bidding aims for a stable cost level, rather than the lowest one, and uses probabilistic pacing. This pacing method selects which auctions to enter based on the likely cost of the optimisation event, and cost levels fluctuate both above and below the target cost.

Bid_Pacing_2018-03

In practice, budget and bid pacing occur as part of the same process. The budget and time left for an ad set affect how the bid is adjusted and which auctions are entered, and the price levels in auctions affect how much budget is being spent. For this reason, you should only control either the bid or the budget, and let Facebook algorithms optimise the other. Once your optimisation and automation are in place, you can make a larger impact on your CPA and ROAS levels by focusing on discovering those combinations of audiences and creatives that perform the best.

Any manual changes done to bids and budgets also affect the learning process for the pacing algorithm, slowing down the discovery of optimal bid levels. The best way to see if your actions affect campaign performance is to set up a proper statistical test and to allow it to gather enough data to reach statistical significance. (Smartly.io customers can check our extensive Support Center for more about ad studies.) 

 

Best practices for setting bids and budgets

  • Is your campaign bid or budget limited? 

Since budgets and bids are coupled through Facebook’s pacing mechanism, you should decide whether it's more important to reach a certain cost level with a flexible budget OR to spend a specific budget completely with flexible cost levels.

The first scenario is called being bid limited (since bid levels are connected to how much you end up paying in an auction and thus define cost levels), and the second one being budget limited. It’s good to ensure that you are setting only one limit (bid or budget) and allowing optimisation to work on the second parameter (budget or bid) to get optimal delivery. Avoid making continuous daily changes to bid or budget levels, as these make it harder for Facebook to optimize campaign delivery.

  • Budget limited? Automate budget optimization 

When your ad sets are budget limited and you still want to control costs, you can use Smartly.io's Predictive Budget Allocation to allocate your campaign budget to ad sets while minimizing the expected campaign level cost per action. The AI behind Predictive Budget Allocation uses Bayesian statistics to predict CPA distributions based on campaign data from the past 14 days, giving more emphasis to more recent data.

You can also allocate budgets across multiple campaigns using budget pools, as long as the CPA levels across campaigns are comparable – this means that you should not include e.g. prospecting and retargeting campaigns in the same budget pool.

  •  Use only 1 optimization feature per level in your campaign structure 

Make sure that you are using only one optimization feature per each conceptual level in the campaign structure – here, for example, lowest cost bidding for bids, and Predictive Budget Allocation for budgets. Otherwise the optimization features are not aware of each other and can end up making contradicting changes in the campaign.

  • Bid limited? Lowest cost bidding with a bid cap 

When your ad sets are bid limited and you want to get most delivery with a given CPA or ROAS goal, as a general rule, you should use lowest cost bidding with a bid cap.

If you also have a specific CPA target to maintain, you can use Predictive Budget Allocation with budget scaling combined with lowest cost bidding. This feature will automatically adjust your budget based on the minimum/maximum budget and CPA targets you have set, while each ad set strives to get the best CPA individually. The Predictive Budget Allocation AI also takes into consideration performance differences between different weekdays (i.e. weekly seasonality) on the level of the overall budget. 

  • Leave room for Facebook's optimization algorithms

Whichever limitation you have chosen, remember to give room for Facebook's optimisation algorithms on the other side of the coupling – for example, if you are budget limited and use lowest cost bidding with a bid cap, use a high enough bid cap to ensure that your budget is used fully, and if you are bid limited, set the budget high enough to get the most delivery at your cost target.

 

Up Next