Data and metrics are what we live and die by in this industry. Yet all too often the metrics we use are over inflated or erroneous. Better numbers tell a better story, but stories are mere fantasies if they are founded on inaccurate data. Many prefer to ignore known data inaccuracies to suit a self-fulfilling agenda. These can range from wanting to sell more media inventory, wanting to prove greater efficiency of a media buy, or wanting to give Wall Street peace of mind that your company is growing. As media practitioners we all have a duty to trade in the currency of facts, not stories. Below I detail 5 key metrics that we use every day to analyze media performance, why they are often inaccurate, and the metrics we should be using in their place to understand facts and not tell stories.
This is the main currency in which digital media is bought, and yet it is the one that comes under the most scrutiny. Most impressions are not seen by the user, they are not even in view, and for the medium term there are no technical solutions to tell you exactly how many people have actually seen your ads. This in turn makes reach and frequency data useless. Yet this is how we buy media and measure online awareness. In the past I have triumphed the viewable impression and warned of the perils of correlating view data with performance metrics. I’m at the stage now where I largely ignore impression data outside of media owner reconciliation. A million impressions minus those that have been ignored, not seen or not served could actually be a hundred thousand (if not less). This makes it a rather pointless and redundant metric.
What you should be analyzing: Unique Views, Viewable Impressions
Whilst impression data is highly unreliable, click data is much more tangible, being an active rather than passive metric. However, a click means someone has interacted with an ad, it does not mean they have deliberately done so. I’ve found in many circumstances that destination page visits are less than half of click volumes. The less qualified or targeted the media, the higher the drop-off from click to destination page is likely to be. Paid search for example, which is highly targeted, can have a drop-off range from 5 to 10%, whilst ad exchanges can have a drop-off of up to 70%, and mobile devices upwards of 80% due to ‘fat finger taps’.
What you should be analyzing: destination page visits
Click through rate therefore becomes an entirely redundant metric when impression and click data are both over inflated and inaccurate. A much more useful metric is the destination page conversion rate. This tells you the number of users who have clicked on an ad and visited your website. This should also be looked in conjunction with destination page dwell time. Low average dwell times and high bounce rates can indicate that there is a high volume of users coming to your website by mistake. Again, highly targeted media normally drives high dwell times and lower bounce rates. I find that average dwell times should exceed a minute to suggest a high level of quality traffic. The next time someone boasts that CTRs are high, I dare you to ask them if they can tell you the ‘deliberate click’ to ‘viewable impression’ percentage – watch them roll their eyes.
What you should be analyzing: Destination Page Visit Rate, Destination Page Dwell Time
Conversion data is often the main KPI for direct response planners. In reality the conversion numbers we receive from our analytics and media dashboards are much higher than true sales figures. This can be due to cancellations and fraudulent purchase activity or because the product being sold requires additional checks (i.e. loan applications) which can cause declines further down the purchase funnel. Having worked on mortgage products the application in principle (AIP) to confirmed mortgage drop-off can be very significant. In online retail some websites have a 40% return rate; your media could be almost half as efficient as you think it is. Most clients I work with can provide daily counts of true sales figures which can easily be matched to your own data. Conversion data should always be reconciled with actual data to give a clear picture of media efficiency.
What you should be analyzing: Actual sales or sales that exclude returns and cancellations
Having worked with many online retailers, the revised adage “high basket value is vanity, profit is sanity” couldn’t be more true. Many online retailers sell products which have little to no margin. This is especially true in the white goods and grocery sectors which are littered with loss leader products. Basket value so often doesn’t give you the full picture. Unless you know the margin of the product you’re selling you’re completely in the dark on true ROI. All too often the client will not divulge this information which is why so many get their agencies to work to a blended CPA target that accounts for low margin goods.
What you should be analyzing: Gross Margin or CPA
Every one of these metrics is flawed to varying degrees, and some of them have much more accurate substitutes which are rarely used. You might not make any friends by highlighting these flaws within your organization, but you might stand out for being a fact finder, rather than a story teller. Longer term this will always serve you well.