Why is the search engine serving that ad?!?

Seeing your ads in some strange places? In response to some unrelated searches? On search phrases you wouldn’t and shouldn’t advertise on? On your competitor’s trademarks even though you don’t advertise on their marks? Seeing the wrong ad served when you have a much more targeted ad that wasn’t served?

Welcome to broad match. It’s getting broader, and it’s forcing smart marketers to change tactics.

Consider the following SERP from MSN (click image to enlarge)

Notice that the ad in position 5 (third down on the right) is for car parts. The search term was “fish”.

The explanation from our MSN rep as to how this could happen actually sounded like this:

Well ‘fish’ could be linked to ‘bass’, and ‘bass’ is not only a type of fish, but a part of a car speaker, and car speakers are car parts…

Ugh!

In the past, the positioning and ad choice was determined by the bid times the click-through-rate (cum Quality Score) chosen from the most relevant keywords. The engines, particularly Google and MSN have learned that Bid times CTR maximizes profits per impression, so who needs relevancy?

This impacts PPC performance in three ways:

1. The retailer pays more per click than they could;
2. they get less traffic than they could because the ad copy is less targeted and click-through rates will be lower; and
3. the conversion rate of that traffic is worse because the landing pages are not as well chosen.

Good for the engines in the short run, but bad for retailers and bad for the engines in the long run as well. Poor conversion rates mean dissatisfied users who are less likely to use the sponsored links next time.

How big is the impact? Let’s look at some numbers:

For terms running on “broad match” some of the searches are actually “exact matches” of the ad phrase, others are “phrase matches”, and some are “broad matches”, in other words broad match ads capture the spectrum. We found that the conversion rates of theses different parts of the spectrum vary widely.

Across all engines for terms on broad match, the traffic that exactly matches the advertised phrase comprises 33% of the click traffic, but 50% of the sales. The rest of the traffic that didn’t match the ad phrase exactly constituted 67% of the traffic and 50% of the sales.

The value of exact match traffic is twice that of more broadly matched traffic.

Moreover, we found that on Google, 25% of traffic coming in on “broad matches” on one term were exact matches of other terms in our campaigns. By swapping out the more targeted ad for the less targeted ad, Google negatively impacts the program.

A smart bidding algorithm with sufficient data will bid more on the more targeted terms and less on the less targeted terms, thus encouraging the engines to “do the right thing.”

However, many of our clients have certain keywords which they want pushed towards the top of the page for branding purposes regardless of their performance, and it is these general terms that become most problematic.

Here’s another interesting phenomena we noticed when sifting through our query logs: people searched for our clients on their trademarks but were served non-brand general category ads instead of the brand ad. We keep the brand bids down to prevent unnecessary spending, but Google found a higher bid to use.

Not only does this increase advertising costs, as we all know, this traffic converts at 5 to 10 times the normal rate. So, stealing some of that super-rich brand traffic to boost the “performance” of the competitive search terms causes over bidding on those terms later…

Probably an unintentional effect…probably.

My tip of the day for PPC advertisers: segment your traffic by match type, watch the performance and bid accordingly, and watch out for cross cannibalization between brand and non-brand ads.

The space continues to get trickier, always keep your eye on the details!