Keyword Decisions: When to Hold On and When to Cut

May 29, 2018

This is a guest post from Oscar Chow, Senior Paid Search Analyst at Wheelhouse DMG.

An ongoing challenge for digital marketers is managing their paid search keywords as efficiently as possible. With the oldest accounts reaching two decades old, it’s not uncommon to find campaigns with unwieldy structures expanding over time. In this post, we’ll go over techniques that can help you optimize your keyword lists for high performance.

Give It Your Best Shot


Let's step away from marketing for a moment and share some words of wisdom from a branch manager from "The Office" who advises, "You miss 100% of the shots you don't take."



In paid search, keywords are the shots that digital marketers take. If a keyword isn’t being bid on but has relevance and conversion potential, that’s a missed opportunity. Much like a star athlete who creates the best chances for their team to score, a fundamentally sound SEM campaign will bid on keywords that have compelling ad copy matched with a high-quality landing page to generate results.

In SEM, having the right keyword in auction at the right time is still a key element to success, even with new search ad formats and campaigns assisted with machine learning becoming more commonplace. Should we keep spending on this keyword that hasn't converted? How much longer can we wait? Can spend be more productive on this group of keywords?

We've heard real concerns like this from many of our clients at one point or another. Rightfully so, the keyword list is often an area where many apply a fine-toothed comb to find opportunities that could spur growth or create efficiencies.

The Process


Let’s take a look at how we can make sense of keyword performance. To aid in the analysis, we’ll address an approach that makes headway but does have a few lurking pitfalls to be wary of. Let's look at an example.

We pull a performance report that screens out keywords that have charged spend over the past 30 days but no conversions. We can title the spreadsheet ‘Inefficient Keywords’.

For starters, we can safely mark (and later remove) keywords that went by previously unnoticed with really high spend and traffic. Obviously, these are poor performers.

As for the others with traffic here and there, doesn’t it seem prudent to remove these "bad" terms and send them to keyword purgatory as well? After all, when you add it all up, it’s a non-trivial amount of spend that didn’t lead to conversion.



But let’s be careful here—by coming to that conclusion, we’re effectively creating a low-light reel. It’s guaranteed that we’ll be disappointed from cherry picking undesirable parts of the data.

Still, it might feel compelling to X-out these keywords. However, because of low sample size and statistical noise, swinging the axe could be premature without further consideration.

To gut-check this reaction, ask the following questions:

1. Has the keyword been given enough of a chance (clicks) to perform?

“Enough of a chance” is going to vary by industry CPC and account. As a rule of thumb, the lower the value per conversion, the lower the spend tolerance should be and vice versa. Stop if the answer is "yes" on this question. You have the confidence to deem the keyword unfit.

2. If there hasn't been sufficient traffic, does performance look any better over a longer period of time?

Time doesn't only heal all wounds—a conversion could be tucked away right outside your set time frame. While you shouldn’t make exceptions common, it’s probably okay to keep a keyword that’s been productive in the past, especially if the window is arbitrary.

3. Would a few conversions drastically improve the keyword's results?

There might be a few keywords on the cusp that would go from stinker to star with just two to three conversions. Keywords with low sample sizes often see high volatility over short time frames because one conversion can drastically skew their conversion rate.

Troubleshooting through these questions should inform whether keeping a keyword active or paused is the right move.

The Solution


To make this type of keyword analysis more scalable, Marin comes built-in with a proactive solution to managing low volume terms—the Dimensions tool. Dimensions allow campaigns and ad groups to be categorized based on intent. The dimensions can span multiple campaigns, allowing for more data aggregation.

Great categorization has three elements:

  • Consistent definitions
  • A depth of attributes for each dimension
  • A breadth of dimensions that encompass the account


Marin’s offering provides all three. Adding descriptive meta-data allows you to cluster low-volume data points into a more representative group, providing a powerful way to make better decisions about keywords on a programmatic scale. In simple terms, we can still make smart decisions with less information than we’d ideally have. Here’s an example of ways you can categorize a long tail, typically low-volume keyword into dimensions to provide more clarity around its value:



Grouping keywords into meaningful clusters, these customizable, client-specific dimensions give us more data to judge. The keyword [garden pruners with 1 inch cutting capacity] may have only spent $10 and not driven any orders over the past 30 days, making it difficult to value. However, if we look at other pruner keywords with no brand name, that have intent for use at home, we can aggregate much more data and assign a relevant bid to this very specific long tail keyword.

Closing Thoughts


The strongest SEM keyword campaigns use practices that put keywords in positions to succeed. You should frequently evaluate your keywords for their ability to be productive at driving results. From time to time, making decisions at the keyword level is warranted to maintain or achieve peak performance. These decisions are best backed by data-driven practices.

Oscar Chow

Wheelhouse
By submitting this form, I am agreeing to Marin’s privacy policy.

See why brands have relied on Marin to manage over $48 billion in spend

No items found.