Yield optimisation is a skill buyers apply on technology platforms to ensure the best possible performance for campaigns. This enables them to optimise the reach and success of their campaigns, in other words, to spend money on inventory that performs well, and have a high chance of conversion.
No ad to show here.
On the seller side, publishers can define certain parameters which ensure that inventory sells at an appropriate (and usually higher) Cost per Thousand (CPM) rate to buyers and that indirect inventory which serves specific campaign objectives is sold at a reasonable selling rate.
For buyers, yield optimisation could be as simple as tracking the click-through-rate (CTR) for a given site and optimizing creatives based on it. On the other hand it could be as complex as setting a variety of buying strategies and campaign specific parameters like time of the day, publisher, ad size, geographical location, channel and price.
All in all, it’s a very important tool for both publishers and advertisers; with this in mind here are some key concepts to get started with yield optimisation.
Understanding yield optimisation
There are a number of manual optimisation levers, which are explained below, as well as automated optimisation of the actual bidding process while the campaign is live.
When using a Demand-Side Platform (DSP), typically every campaign starts in a “learn” phase, and then moves into an “optimised” phase once it has gathered enough data. At the start of a campaign, every bid on inventory is a learning bid, and the optimisation engine attempts to determine the best price to spend on each piece of inventory based on the performance of similar inventory in previous campaigns and is measured against the set of parameters that are determined at the onset.
When a defined number of success events -called the threshold – are reached on each node, more inventory is optimised and the campaign has the opportunity to spend more of its budget in the optimised phase.
When you load a new campaign, performance data is not yet available and the optimisation engine must first learn what to bid for the campaign. As the campaign wins impressions, the algorithm takes new data into account, causing learn bids to increase on well performing slices of inventory and to decrease on poor performing venues. In the case of repeat campaigns, the optimisation engine uses a campaign’s past performance data to determine the appropriate bid for every impression relative to the campaign’s cost-per-acquisition (CPA) or cost-per-click (CPC) goal.
What you need to know
1. It’s vital in an RTB environment
This is because in an exchange, advertisers are predominantly performance driven and hunting for the best possible CTR or CPA, so it makes sense for you to know what the probability is of a certain impression doing well for your campaign or not.
2. Publisher focused yield optimisation
From a publisher’s perspective, optimisation starts with describing, classifying and certifying inventory so that it’s easily verifiable and buyable across the network of Demand-Side Platform’s (DSP’s), exchanges and advertisers. For publishers, it’s also important to understand that each publisher can configure a unique ad quality profile that ensures that there is no conflict between their direct and indirect sales channels. This can be done by:
- Creating a block list of specific brands or advertisers that a publisher may want to keep within their direct sales channel.
- Specific content categories that may be in conflict with editorial policies.
- Technical attributes that prevent certain creative formats from being delivered.
Quality profiles ensure that the publisher protects its creative and editorial integrity.
The next step is for publishers to establish price floors for the different types of inventory, which should be set in conjunction with the sales team to merge with their direct sales strategy. This can also be influenced by the quality of the publisher, ad sizes and positions, ad quality guidelines and creative rules.
Once publishers go live, price floors are continuously adjusted based on reporting data. As part of their yield optimisation strategy, publishers can also create audience segments of their inventory to sell at higher CPM rates to specific buyers and also create packages of specific groups of inventory in the marketplace to sell to all other buyers. These packages can then be further modified and converted into deals to sell to the DSP. Once there is enough demand for a publisher’s inventory, they can start building out private exchanges, where they would make certain sections of inventory available to a select number of buyers – based on certain criteria. These criteria could, for example, include a minimum monthly spend with the direct sales channel before unlocking access to certain programmatic inventory, at a reduced price.
3. Advertiser focused yield optimisation
For a buyer, assuming they have no pre-existing data to work from, the optimisation process starts with their choice of buying strategy. There are a number of buying strategies but some of the most common include:
- Simply buying across the network – possibly for brand awareness or to build a remarketing list for later use.
- Target specific interest categories websites or audience segments.
- Or buyers might decide to drill down to the placements that perform best for a given brand category.
Once the strategy has been decided, buyers can input their desired bid rates and set performance goals with regard to CPC, CTR or CPA which allows the buyer to track actual performance, versus desired performance. Further optimisation allows buyers to set target CPC’s or CTR’s, above which they would not bid for in an auction environment.
4. Publisher yield parameters
The key yield parameters for publishers are:
- Making sure that the programmatic offering doesn’t undermine their direct sales efforts.
- Making sure that inventory is visible and available to buyers in an open market.
- Price floors are realistic.
A big part of optimisation for publishers is around how the inventory is packaged and how it is made available – not only as clearly defined placements, but also as desirable audience segments such as job seekers, technology enthusiasts or car buyers.
5. Advertiser yield parameters
From a buyer’s perspective, there are a number of parameters that can be leveraged to refine their audience targeting. On a simple level, this can include:
- Setting CPM bid rates.
- Applying frequency caps.
- Geo-targeting.
- Operating system and browser targeting.
- Day and time targeting.
On an audience level, it’s possible to target websites, or more granularly to specific website placements or universal or custom content categories – where you can buy against an interest group rather than an audience.
But perhaps the most valuable optimisation lever available for advertisers is their ability to create and continuously refine their own audience segments, which can then be used for future remarketing campaigns. A good example of this would be to create segments of users who have shown intent, and segments of users who have actually converted – and then excluding the converted user segment from future campaigns – why show someone a banner when they have already bought your product? If we drill down further, knowing that they have bought your product, you can then choose to show them a campaign for a complementary product or offering – further increasing the probability of a higher conversion rate.
The power of optimisation
Optimizing campaigns and inventory can be a time consuming and tedious task. But all the tweaking and changes are well worth it in the end – it can create hugely successful campaigns for advertisers, and increase the overall bottom line for publishers. First understand optimisation, identify your parameters in line with your overall strategy, and then continuously update them to ensure that you harness the full power of yield optimisation.
