Analysing your search terms, and acting on the data they show, is one of the core optimisation activities for search campaigns.
That’s pretty uncontroversial, unlike Google’s recent move to reduce the amount of search term data available to advertisers.
With less search term data to work with, we definitely want to make the best use of what we have.
Here are three tips to help make sure we’re doing that…
1) Add the keyword column
When scanning search terms within an ad group, it’s useful to add the column for ‘keyword’, showing which of your keywords triggered each search term, and the resulting data.
This gives you much clearer insight into how your keywords are operating, and whether you’re seeing search term overlap.
It’s also an interesting window into how different match types are working in practice.
2) Act on Patterns
Bad Search Terms (BSTs) tend to hunt in packs…
It’s usually better not simply to hit the ‘add as negative’ button when you find one. These negatives would then only catch the specific phrase in question, which may rarely or never show up again in any case…
Instead choose a negative to target the broadest (and shortest) element that makes the phrase irrelevant, catching similar BSTs in its net, but leaving all of your desired traffic intact.
For example, say you’re advertising kitchens…
You see the search term: ‘wooden kitchen set for kids’… Rather than turning that long-tail search phrase into a negative, set a wider net with the negative kids… and throw in child, childs, children, childrens while you’re at it, to eliminate the whole playset / toy (let’s add those too) theme.
Then there are the already-visible patterns among your existing BSTs: phrases, or n-grams, that show up repeatedly within them.
An informal way to identify these is to scan your top-spending, non-converting search terms for common words or phrases… and test their value as a whole by running a filter for all search terms including [suspect phrase].
If those filtered terms are performing substantially worse than your average, you have enough evidence to convict them.
The more formal way is to use an n-grams script that scans for all relevant patterns of 1, 2, 3 or 4 consecutive words among your search terms, and their associated performance. It’s powerful, but quite heavy-duty.
For the same output without the difficulty and time-suck of working with scripts, Opteo* runs this analysis by default.
3) Added/Excluded Analysis
The Added/Excluded column shows whether each search term you’re evaluating is an exact match for one of your keywords – and whether it’s already excluded with a (exactly matching) negative. Important context for your work within the search term report.
But this variable can also be used for a nice piece of account-level analysis.
If you head over to Google Ads reporting section, and create a table with Added/Excluded as your primary variable… add in impressions, clicks, cost, and conversions (we’ll work with conversions rather than value here to keep it simpler).
Check the output, and you’ll see…
A very unpromising start. Don’t worry, we’re not done 😁.
This report won’t aggregate your keywords by their Added/Excluded status, but we can sort that with a pivot table.
- Export the results into a csv, and upload it to Google Sheets
- Kill the first two rows
- Create a pivot table (by selecting Data > Pivot Table > New Sheet)
- Choose Added/Excluded as the row, and your Google Ads metrics as the values, you’ll see the data as aggregated across your account
The key analysis here is the difference in impressions between ‘added’ (i.e. search term is an exact match for a keyword) and ‘none’, which implies some degree of search term ‘expansion’ from your specified keyword.
If you’re using any match type other than exact, you’re inviting this expansion, which is entirely valid, but remember that the further your search terms stray from your specified keywords, the less relevant – and precise – your keywords become as units of optimisation.
So in short, if ‘none’ accounts for more than around 80% – you might want to look at pruning your search terms more vigorously with negatives, and refining your match types. (There’s plenty of room to quibble over the threshold for a healthy % here… but the higher it is, the less control you have over which specific search terms are triggering your ads, and what ads they’re triggering.)
If you shift your gaze over to conversions and cost, you’ll see how much of each is coming from these less ‘controlled’ impressions.
NB this data only covers your visible search terms, and the insight is slightly different from (but in the same bracket as) the Lin Rodnitzky Ratio.
As someone sharp-eyed in my Facebook group recently pointed out, the search term cull compromises the simple L-R ratio formula… but the analysis above can give you a similar overview of how tightly-focussed an account’s spend is.
Finally, it’s worth noting that – at time of writing – you can still see most of those missing search terms in Google Analytics (Acquisition > Google Ads > Search Queries) but that’s not likely to continue…
For now, we’ll have to swallow the recent cull in search term data, so let’s make the most of what we have.
Want to go deeper into keywords?
- Keyword Research
- Keyword-Level Bidding
- Search Terms
- Negatives
- Match Types
- Quality Score
