Single Request Architecture Example

After you've integrated the Google Publisher Tag (GPT) library into your site and become familiar with our [general best practices](https://developers.google.com/publisher-tag/guides/general-best-practices), you're ready to start issuing ad requests. In the following sections, we'll cover some additional best practices you should keep in mind when configuring and working with ad slots, to make the most of your ad space with minimal performance impact. ## Prioritize "important" ad slots Not all ad slots are created equal. For example, slots which are visible as soon as your page loads (above the fold) are usually more "important" than those which are not visible until scrolled into view (below the fold), in terms of viewability and monetization. With this in mind, it's important that you carefully consider the relative importance of every ad slot on your page and prioritize loading the most important slots as quickly as possible. ### Load above the fold ads early Ads that will be visible as soon as the page loads should be given the highest priority. It's recommended that you define these slots in the `` of your document, and request them as early in the page load process as possible. This helps to ensure that these ads are loaded early (maximizing viewability) and that they won't unnecessarily slow down the initial page load. ### Load below the fold ads lazily For ads that need to be scrolled into view, fetching and rendering should be deferred until the slots are close to entering the viewport. This is a process known as lazy loading. [Lazy loading](https://developers.google.com/publisher-tag/samples/lazy-loading) separately prioritizes requesting and rendering creative content for slots that are the most likely to be viewed. This helps optimize page load performance by conserving the browser's limited resources, which is especially important in mobile environments where bandwidth and CPU are often heavily constrained. ## Refresh ads without refreshing the page There are many scenarios in which it is optimal or even necessary to replace the current ad content of a slot. In these cases, it is best to use the GPT library's [refresh functionality](https://developers.google.com/publisher-tag/guides/control-ad-loading#refresh) to do so dynamically. This avoids a full page refresh and lets you precisely control the conditions under which a slot or group of slots are updated. When refreshing ad slots, it's important to be familiar with and adhere to [`refresh()` best practices](https://developers.google.com/publisher-tag/guides/control-ad-loading#best_practices). Refreshing ads inappropriately can lead to performance issues and negatively impact viewability rates. > [!IMPORTANT] > **Important:** To comply with Google policy and enable your inventory to compete on Ad Exchange, you must [declare which portions of your inventory refresh](https://support.google.com/admanager/answer/6286179). ## Target ads effectively When configuring [key-value targeting](https://developers.google.com/publisher-tag/guides/key-value-targeting), carefully consider whether to use slot- or page-level targeting. For key-values shared between multiple slots, it's most effective to use page-level targeting using [`googletag.setConfig({ targeting: ... })`](https://developers.google.com/publisher-tag/reference#googletag.config.PageSettingsConfig.targeting). Slot-level targeting should only be used to register key-values which differ or are not included on all slots. Note that both slot- and page-level targeting can be used at once, as seen in the [set targeting example](https://developers.google.com/publisher-tag/guides/key-value-targeting#set_targeting). It is strongly recommended that you configure targeting at the page-level first, then apply slot-level overrides only where necessary. This approach not only makes efficient use of the GPT API, it also simplifies code and helps in maintaining a clear mental model of all targeting configured on the page. ## Use Single Request Architecture correctly Single Request Architecture (SRA) is a GPT request mode which bundles requests for multiple ad slots into a single ad request. This ensures that competitive exclusions and roadblocks configured for your page will be honored. Therefore, if your page makes use of these, it's recommended that you enable SRA and understand how to use it correctly. In its default configuration, SRA will request all ad slots defined on your page the first time you call [`display()`](https://developers.google.com/publisher-tag/reference#googletag.display) (or `refresh()`, if [initial load is disabled](https://developers.google.com/publisher-tag/guides/control-ad-loading)). Because of this it's recommended that you define all of your page's ad slots in the `` of the document, before making the first call to `display()`. [Lazy loading](https://developers.google.com/publisher-tag/samples/lazy-loading) can be used in conjunction with this approach to ensure slots located below the fold are [not loaded immediately](https://developers.google.com/publisher-tag/guides/ad-best-practices#below-the-fold). When using SRA, it's important to fully configure all ad slots (for example, set targeting, category exclusion, etc.) prior to making the first call to `display()`. Only values configured before this point will be included in the initial SRA request. Incorrect --- ad slot configuration is not included in SRA request Single Request Architecture Example Correct --- ad slot configuration is included in SRA request Single Request Architecture Example ## Optimize your ad sizing When defining your ad slots, consider not just the maximum size ad that could serve, but also smaller sizes that could comfortably fit into the same space. In general, the more sizes you specify when defining a slot, the more ads that will be eligible to serve to it. This can translate to higher fill rates and increased revenue.