Kevin Anchi

Google just announced they are taking near match out of beta and rolling it out to everyone.

What this does is match you to close variations of search queries. It’s pretty similar to modified broad match, only it works for phrase and exact match.

What Near Match Does

For instance, if you are advertising on the keyword: “buy plasma TV”; you will NOT show for these queries:

• buying plasma TV (stemming)
• buy plasma TVs (plural)
• buy plasa TV (misspelling)

With the near match options, you will show for those keywords.

Dartboard Image credit: Velo Steve

Google Does It Again – Auto Opts You In – You Have to Opt Out

I really don’t like when Google does this – they opt you into something instead of leaving existing campaigns alone. I logged into an account today and all the campaigns were changed to the new match type by default.

So, if you don’t want to use it, you need to go through every campaign and disable it. It’s not in the AdWords editor yet, so you have to do this campaign by campaign. You do have until mid-May (supposedly) to change this.

To change the settings; just go to the campaign settings, and near the bottom of the page you’ll see the Exact and phrase matching options:

From there you can turn it off (or enable it if you turned it off earlier).

Update:  Various readers are seeing a large variety of things going on in their accounts. Some are seeing that all campaigns were opted in, and now only half are opted in. Others are seeing that they are completely opted in, yet others are changing the setting only to have it changed again after a couple hours (without them doing anything).

It seems that Google should not have launched a setting that wasn’t in use this far in advance of it going live. It’s only making people more confused and annoyed.

Will Exact Match Trump Near Exact Match?

I’ve talked to several people at Google about the near match ad rank issues to see what will trump what. I don’t have a clear answer. This is what I hear:

• Google always uses the most precise matching option, so if you have the exact match version of the word and the near exact match version of the word; then the exact match will be displayed
  • This isn’t true to begin with (always uses most precise match); so the extrapolated answer seems incorrect; but it could be correct
• Ad Rank is max CPC x QS. Since QS will be the same, then if your near exact match bid keyword is much higher than the exact match version of the word, then the near exact match will show and not the exact match
  • I’m guessing this is correct; and it could mess up stats.

How Stats Can Become Corrupted

Let’s say you have these two words:

• [Restaurant waiter clog]
• [Restaurant waiter clogs]

There are usually less ads on the ‘clog’ version of this keyword, and the CPC is generally lower than the ‘clogs’ version. However, it also has a lower conversion rate. Therefore, you bid the ‘clog’ version down a little bit, but as there are so few ads, it doesn’t matter.

Now, with everyone being opted in by default into this option, the ‘clog’ version is going to get a lot more competitive; so the CPCs will go up. As the ‘clog’ version has a lower conversion rate, you just accept the fact that you’ll get fewer conversions from this word and bid the word based upon ROAS.

However, the bid is so low for the ‘clog’ version that when someone searches for ‘restaurant waiter clog’ Google no longer triggers the exact match version; they trigger the ‘clogs’ version as it is a ‘near exact match’.

Now, the search query report should show that the ‘clog’ version received the click; but you can’t bid on a search query. You have to add it as a match type first. But, this keyword is already a match type with a lower CPC. So, you’ve now lost ad serving control.

How Often Do “Near Exacts” Have Different Conversion Rates?

That really depends (yeah, I hate the answer too).

I’ve dealt with accounts where the singular and plural versions sold different products, or used different pages.

I’ve dealt with accounts where everything similar behaves the exact same way.

Only you can find out this data for yourself. Take a look through the search query report and see if there are any commonalities amount singular words, plural words, misspellings, and stemmings.

It’s the stemmings I’m more concerned with that the plurals or misspellings.

What will this be matched to (note: phrase match): “Cleveland Driver”

• Cleveland driver (brand of golf club)
• Cleveland drive (the 1987 John Elway Game ‘The Drive’) or a street name
• Cleveland driving school
• Cleveland drivers license (drivers vs driver is a huge difference here)

And I’m sure there are much better examples that will come to those who have had more coffee than myself.

Can You Test It?

You can test almost anything; but this will be really hard to test. Cross campaign ACE (AdWords campaign experiments) would be really useful. That is among my top 5 wish list items for AdWords.

From a conceptual standpoint, there seem to be two ways to test it.

Exact Match Positive & Negative Keywords

Sometimes, Google won’t show you when you have the same word as a positive and negative in the same ad group or campaign – so this might not work.

• Copy/paste your exact match keywords to a new campaign
• Copy/paste those same keywords as negative exact match keywords
• Enable the setting
• See if you get traffic (uncertain if you will)

Since the new campaign’s keywords say show me for this, but don’t show me for the exact same item – then only if it’s a near match should the keyword be displayed.

In the old campaign, leave the ‘near match’ setting off so that it will capture the true exact match data.

Collect some info and compare the two campaigns.

Duplicate Campaigns w/ Lower Bids

This method is how you use to control search partners:

• Duplicate the campaign
• Change the setting to ‘near match’
• Lower the bids by 10%-20%
• Collect the data
• Examine the results

Now, this method is not as good as the other one. As the bids are lower, you will receive fewer clicks. And as you’re not controlling the info with negatives, you’ll get some corrupted data. However, it’s much faster than trying to match up all the negatives.

Should You Use It?

I have very mixed feelings about near match.

I work with some accounts that have loaded up on so many exact match and phrase match variations of words for their auto-bidding system; that the setting really isn’t that useful. In this case test it or leave it off.

I work with some accounts that have loaded up on exacts and phrases when it seemed useful, but they don’t spent enough time really controlling all of the display and they prefer to let the bidding system handle it. In this case, test out the setting to see how it performs.

In some accounts modified broad match is doing great, and when there’s a lot of traffic they also add the keyword as an exact match. For these accounts, it might be more useful to let the modified broad catch the ‘near match’ impressions and then use search queries, negatives, and true exacts to manage the bids – when there is a lot of traffic.

I think that might be the difference, the above management method is great for medium to large accounts. It is very difficult for accounts with little traffic.

So… are you a control freak? Then leave it off or split it out into another campaign.

Do you want the most exposure for the least amount of work, or do you have a small data set? If so, turn it on.

Are you in the medical jargon industry where misspellings outnumber correct ones? If so, turn it on.

Time vs Control

With these new betas, such as Display Campaign Optimizer, it comes down to control vs time. If you want lots of control, then it’ll take you longer, but you have control.

If you don’t want a lot of control, or have very little time, turn them on, benchmark how they do, and then decide to keep or turn off the setting.

As Google tries to make AdWords more appealing to the masses who don’t have lots of time, or sophistication; expect to see even more features that offer time saving methods, but at the cost of losing some control.