Yield Optimization for First-Price Auction

Simplicity and transparency will never happen.

In theory, first-price auctions bring simplicity and transparency. The model assumes that buyers will bid exactly what they perceive as a value of the impression and everyone will be satisfied. The problem is that it will never happen, because, in AdTech, people love to optimize. Whatever auction model you work with, two things never change: buyers want to spend less, and sellers want to earn more. The conflict between them is an engine that drives AdTech development.

Why yield optimization will be harder?

First-price yield optimization goes long for the sell-side. It is still a game between sellers and buyers, but only buyers see results immediately.

In second-price auctions, sellers can look for the best revenue by managing floor prices. They can even split the traffic for equal parts and apply different pricing to see how prices are determining total revenue, but with first-price, each floor-price decrease increases revenues in the short term and vice versa. Each time floor prices go up, revenue goes down in the short run.

So if you think about tomorrow’s revenue, the best thing is to set the floor to 0 but remember, buyers, are analyzing your prices in an automated way, looking for the lowest price for which they can still buy your inventory. The cost of tomorrow’s revenue maximization is a risk of getting lower bids later on and having a revenue reduction.

Also, header bidding will not push your Google AdX any more. It will still have value for you, especially when your header partners have unique demand, and the margins they have are lower than Google margins.

Do you like what you’re reading? Subscribe our newsletter for more content like this!

[mc4wp_form id=”407″]

What will the change look like?

Keep calm. Things will change, but it is quite easy to not lose after this switch.

You can even increase your revenue. This change will not increase or decrease the value of global ad spends in a meaningful way, so there will still be enough money on the market, just don’t let others take a part of your share. Keep in mind that in the beginning, your revenue will probably rise due to some buyers that will be late with buying-strategy adoption and will still bid as they would in the second-price world.

With time, the revenue will probably drop, as machines on the buy-side will be looking for the lowest possible price for which they can buy inventory.

What do we suggest you do, in order to protect your revenue?

Below we present you two sets of tips, you can adapt to your revenue optimization methods.

#1

(The simple version if you’ve never optimized open market).

• Never set your floor price at 0. That was acceptable (well, not for us at Yieldbird) for second-price auctions, but now machines on the buy side will quickly understand that they can bid really low and still buy traffic from you. If that happens, you will probably wait much longer to get paid better.
• Analyze your historical performance and RPM (revenue per 1000 Impressions). Understand the rhythm of the trade. There are better and worse performing days and months. Take that into account when building your pricing.
• Keep pricing consistent. Do not let any header partner sell under the price.
• Forget A/B testing. It’s much harder with first-price auctions.

#2

(The advanced version for hard open-market optimizers and for Yieldbird).

• Understand buyers elasticity and time of reaction. How fast are they adapting their buying strategies to your floor prices?
• Build models that will constantly analyze spending from buyers, their bid landscapes, and buyer-specific price elasticity to understand if you floor them too low, too high, or in the optimal range.
• Build automated models that can implement the process for a whole inventory, taking into account the diversity (different websites, ad units, audiences, and platforms).
• Understand the market. How are your competitors pricing their inventory and how often do they make changes?
• Stop A/B testing, but build a model that you can afford and that can help you evaluate results and make experiments. Apply different pricing strategies for separate ad units or even websites
• Call sellers much more than you have in the past. Ask for the price that they want to spend and open a PMP with them. Many sellers will be happy to have a predictable price instead of playing the game of buying and selling.

By doing this, you will likely be above the market and can increase your revenue by getting part of the share from other, less active publishers and buyers.

Of course, it’s not easy and requires a skilled, multidisciplinary team of ad ops, developers, analytics, data scientists, and sellers.

However, if you are big enough that 20% uplift from your open-market revenue can cover the cost of that team. You can consider it an investment. If not, look for outsourcing or implement the simple version.