Using data injection for the forecast

On your websites, you may have exceptional (temporary) traffic increases or decreases. For instance, traffic may decrease significantly during the Christmas holidays; conversely, traffic may increase during an election period.

Smart AdServer's forecast cannot "know" about such exceptional traffic increases/decreases. It will always assume that the traffic trend will continue as it did in the past.

Here, Smart AdServer's feature data injection comes into play. It allows you to force the forecast to use custom traffic data.

Setting up a data injection

Setting up a data injection

Go to Administration > Data injection in the inventory availability

Under Available placements (1) select the websites/pages/packs, for which you want to apply the data injection and move them to right using the blue arrow icon (2).

Select the Start date (3) end the End date (4) during which the data injection should be used.

For Type select Percentage or Raw value (5)

Percentage: You can use percentage values greater or less than 100% (6). The forecast will apply this percentage to the number of page views (page impressions) of the selected placements.
Example: if you enter 120% for a website with 1 million page views, the forecast will assume that the website will have 1.2 million page views (during the timeframe of the data injection) and will calculate accordingly.
Entering 100% would not make sense, because it would tell the forecast to use the volume of page views it uses anyway.

Raw value: You can also use a raw (absolute) number of page views. Keep in mind that this value is for page views per day (not for the entire timeframe of the data injection). The raw value entered here will replace the value calculated by the Forecast for the number of page views.

When done with the settings, click Save.

On the History tab, you get an overview of existing data injections. You can also edit or delete data injections.
You have to wait until the next day, until newly created data injections or changes to existing data injections will take effect. This applies to both the premium and the free version of the forecast.

Best practices

Overlapping data injections

If there are multiple data injections with overlapping criteria (overlapping timeframe, overlapping websites/pages/packs), the most recently edited (or created) data injection will apply.
We strongly recommend to avoid overlapping criteria. This may lead to behaviors difficult to analyze and fix. The best practice is to split the data injections into multiple data injections without overlaps.

Updating data injections

If your data injection uses percentage values, please check if the percentages you chose were correct a few days after the data injection's start date. If you detect that the actual traffic is much higher or lower than the percentages you defined in the data injection, you should update the percentages. If you leave the data injection unchanged, you risk that the forecast interprets the data injection as a global trend rather than an exceptional (temporary) traffic increase/decrease.

Was this article helpful?
1 out of 1 found this helpful
Powered by Zendesk