Traffic Slashing: The Silent Killer
When managing campaigns, it is tempting to pause off poor-performing keywords. The logic is pretty simple; I only have X dollars so let’s just push the traffic at what is working and pause everything else. This seems intuitive but actually is the wrong way to look at it. For the record, we see this error all the time when we run audits. The main takeaway is that any KW can be a good KW and any KWs can be a bad KW depending on the CPC. If you pause KWs without optimizing then you lose potentially profitable traffic and force higher CPCs to hit the same budget. Let us take this hat store example.
Main Issue
If you look at these 5 KWs you might be tempted to kill anything that has a lower than 1 ROAS and funnel all traffic to the performers (baseball and party hats). Instead, the answer is to find the value of the poor performers and lower your bid until you are able to produce profitable traffic. A keyword that is on with a low bid is more valuable than a paused one, even if the activated keyword spends little or even nothing at times. Every visit has value, the trick is to have a bid that reflects that value. In this case, it still makes sense to cut “party cat” as that is ridiculous and has no rev but “free hat” is potentially still good even though cerebrally we would want to cut it.
Why Bother?
Ok so there you have it. The same amount of spend but more revenue. Win-win. In this example, we are only looking at a handful of KWs so the amount of improved revenue is relatively modest but if we take into account every KW this will make a gigantic difference in performance.
Keywords are not the only example of traffic slashing either. Anytime you pause or exclude a device, geo, or time of day then we are saying that there is 0 value to that traffic. In reality, we mean that they have less revenue per click than average. All we need to do is find a bid that reflects that value and optimize that traffic. This is the same reason why you want to opt into all placements on Facebook and never do blanket exclusions on demographics.
