3 Forecasting Strategies For PPC Data Anomalies

COVID-19 is impacting every area of our lives. Working for an agency, or within a marketing department, I imagine it impacted your client’s marketing budgets (or your own) for the immediate future and most likely, for the rest of the year and into next year. 

I have a client that was severely affected by COVID-19 in an extremely positive way – they could not keep up with the influx of demand. This drastically shifted our digital priorities and budgets. We are now in the midst of putting together budgets for their upcoming fiscal year, and in doing so, ran into quite a few hiccups because of COVID-19 and the resulting performance and budget shifts. 

As a result, I’ve found there are a few different ways to review PPC data from the COVID-19 period.

There are three scenarios I would encourage you to look at as you create new budget projections for this year:

• Digital revenue as is (no adjustments, no normalized numbers, no accounting for increases/decreases in spend)
  • This situation focuses on revenue to dollar ratio that actually happened before, during, and continuing through the virus outbreak
• Digital revenue with the impact of COVID-19
  • This situation focuses on revenue to dollar ratio and any compensations that were made to your marketing budget due to the virus outbreak
  • Situation two is different from situation one as situation two takes into account if you raised or lowered your digital spend due to COVID-19. Situation one focuses on maintaining everything ‘as-is.’
• Digital revenue without the impact of COVID-19
  • The final situation focuses on throwing out data from the COVID-19 period and considering whatever performance happened as an anomaly. A line would have to be drawn at the onset of the virus to an agreed-upon date to when normalcy in performance is/was anticipated.
    • In the case of my client, it was March 15 – to date (or until demand died down to normal levels.)

Strategy One – Projections Without Adjustments

This situation is likely the easiest to account for projections. Many marketers would likely deem this situation ‘passive.’ It is putting you in a passive position where you aren’t letting external factors impact your decisions. Your marketing dollars are set and you will not shift one way or another due to performance. 

To forecast within the parameters of situation one, project budgets with all metrics remaining constant. For example, all CPCs, impression share, CTR, and CVR are assumed to not change compared to your lookback window (lookback window = the time frame you are referencing performance to project future time frame, such as last year, six months, thirty days, etc.). 

In the example below, you’ll see that the revenue forecast is closely correlated to changes in the cost or budget. That is because we assumed all other metrics, like CPCs and impression share, would remain constant.

Strategy Two – Accounting For Fluctuations

Situation two differs from situation one because you are accounting for any fluctuations in the budget you made due to the virus. This situation also anticipates performance taking time before it adjusts whether it is because the virus will calm down, seasonality, or general interest returns or declines. This situation adjusts for seasonalities AND anomalies. 

When accounting for fluctuations due to COVID, or any other unexpected event, you are taking into account the performance from the time period of the event and including that data in your averages (or however you choose to do your forecasting). Depending on the industry, this strategy may be the most misleading in budget forecasts. 

Strategy Three – Excluding The Anomaly

The final situation to consider is likely the most difficult because you will have to choose a date that is your line in the sand when performance shifts due to COVID-19. For example, my client chose March 15. From March 15 on, we can treat performance as an anomaly and, depending on if COVID-19 negatively or positively impacted performance, we assume that we will see similar results from our line in the sand to a designated date set (for the next 2 months, 3 months, 6 months, etc. past your set COVID-19 start date).

How you ultimately calculate situation three depends on how you calculate projections normally. You’re throwing out a significant chunk of time for PPC performance. There are a few ways you can calculate your budget forecasts without the excluded date range:

• Use the average of your performance for the last three months preceding COVID-19 (or any other anomaly) and if possible, combine that with the average for the next three months past your designated COVID-19 end date.
• Keep your costs constant (CPCs) and fluctuate total spend based on your CTR or CVR goals. 
• Keep CTR and/or CVR constant and fluctuate your costs (CPCs and/or total spend) based on your return goals (CPL or ROAS.) 

To fully understand the impact of COVID or another anomaly on future performance, I’d recommend reviewing projections with each of these scenarios. This can help you and your client or department make the most informed decisions.