SEM - Thousands of Keywords - How To Optimize Cheaply?

[Zach]

In the last few months I've spent $10,000+ on a fairly open keyword strategy for our product. Mining for data.

I'm in a place where there's just so many keywords I only have like 10% of them I'm even ready to make a decision on --> ditch or bid up. It's all still just noise.

"Hmmmm....4 clicks on this keyword.".....times 5,000

Do any of you have any good strategies for optimizing before any of these keywords are fully mature with data?

The only strategy I have is to theme out keywords & make them into "families" where the intent seems to be related. Then I make decisions on these "families" as one unit, where the significance of the spend & conversions is worthwhile to look at. I'll keep a "broad beta" ad group out there for continued KW family mining.

I know this is a common scenario for many of you in Adwords.

What do you do when there's just not enough data yet?

Ditto for Google Shopping (where I have MUCH less levers to pull).

 

[eliquid]

Yeah, how you started was the issue.

You never go full retard ( in your case, full shotgun blast with thousands of keywords bidding on ).

 

[ryandiscord]

You may want to restructure. I'd start by looking at what keywords and search terms converted and create related ad groups based on those. Pause everything else. You can slowly reintroduce the other keywords as you test. If you aren't already, consider using modified broad match for data collection.

Depending on how many skus you have will determine the complexity of your shopping campaigns. Generally I would let shopping run if I was seeing positive return and use that to mine data. If it's not profitable, look at your impression share. If you need to, bid lower and capture less impression share while you sort things out.

 

[Zach]

Yeah I started with Google Shopping. Just plugged it into my Shopify store and let it go. As long as it didn't lose money I didn't focus on it much. Just figured I'd loop back after I spent some good cash for data.

It's a 1 product store.

In hindsight, I think I would've liked to START with the standard adwords search campaign, get a really good idea on the customer funnel for AIDA keywords, THEN use that to aggressively optimize the limited-targeting of the Shopping campaign.

 

[ryandiscord]

In hindsight, I think I would've liked to START with the standard adwords search campaign, get a really good idea on the customer funnel for AIDA keywords, THEN use that to aggressively optimize the limited-targeting of the Shopping campaign.

To be honest, that probably wouldn't matter. Google is really good at matching shopping ads with buying intent searches. Pull your search terms report, I bet they're mostly buying intent anyway. That and you can't really keyword target a shopping campaign. They show based on the data in your shopping feed, in your case your Shopify store. You can add negative keywords but that's about it.

 

[CCarter]

As a macro overview - I'm a light-weight on paid, but generally you do the following:

1. Determine the most profitable product/service in your array. This way if you get sales the bottom-line will be impacted faster and you can then somewhat relax to get results for lesser profitable product/services. As well since it's got the largest margin you can be a bit higher and have room to breath.

2. Determine the landing page that converts that product/service. If it's eCommerce it's not hard to figure this out since it's the main product page. If it's a service find out. If you do not have a single page dedicated towards your most profitable product/service - create one.

3. Find 3-4 keywords AT MAXIMUM that have buying intent behind them for the product/service. Going broad is extremely bad. I give an example of the within the Keyword Research of the crash course: (Day 6 - Keyword Research)

"Our focus is on profits - everything we are after is for profits aka the biggest bang for our efforts. Keyword research is the difference between you making 20 sales a day by ranking for 'womens shoes' (Google shopping's organic results values products at $24.95 to $85) - generating between $499 to $1700 in revenue a day, versus making 10 sales a day by ranking for 'red bottom heels with diamonds' which cost $795 to $995 per pair - generating you $7950 to $9950 in revenue per day."​

So using that logic, don't go broad, but go extremely targeted with buying intent behind it like "buy red bottom heels with diamonds". You have to research the terms people in your industry use for buying intent - example "buy", "sign up", "download", etc. You can't "download" a BMW, but you can "get a BMW quote".

4. Double down. Once you start seeing some good conversions, expand upon the successful keywords and remove the one doing horrible. If you have a targeted CPA use that are your metric, otherwise you know your profits and whether there is potential for up-sell or additional revenue down the road from a client, so even a negative CPA might become profitable afterwards.

--

Every campaign should start narrow otherwise you'll get noise from the keywords, the users' intent, the users, the landing page conversion, and all the fun of it all coming together. No one can handle that much data coming at them with a new campaign.

 

[ryandiscord]

3. Find 3-4 keywords AT MAXIMUM that have buying intent behind them for the product/service. Going broad is extremely bad. I give an example of the within the Keyword Research of the crash course: (Day 6 - Keyword Research)

You can reverse this logic a little bit and find phrases with the wrong intent and use them as negatives. For example if I'm selling a new product, I would exclude "used", "rent", "rental" and "free".

 

[Zach]

To be honest, that probably wouldn't matter. Google is really good at matching shopping ads with buying intent searches. Pull your search terms report, I bet they're mostly buying intent anyway. That and you can't really keyword target a shopping campaign. They show based on the data in your shopping feed, in your case your Shopify store. You can add negative keywords but that's about it.

For Google shopping, I've seen people use a 3-tier ISO strategy (utilizing the high/med/low campaign priorities) to "stack" their campaigns by high/med/low bids based on intent. I am interested in replicating the approach, but I can't help but think just using the Google Shopping CPA tool would be 10x easier. But I've also seen many of Google's bidding tools suck really bad too.

Any thoughts on Google's CPA tool for Shopping campaigns?

You can reverse this logic a little bit and find phrases with the wrong intent and use them as negatives. For example if I'm selling a new product, I would exclude "used", "rent", "rental" and "free".

Ya this is the approach I tend to do. I exclude KWs based on "low-intent phrases". However it does have its cons as a few have mentioned already.


In my case for Google Shopping, I've got negatives on stuff that would be informational searches like "how to" "what is" in addition to all the other standard unrelated negative keywords. Google shopping ads just feel really ill-equipped to do well with the high funnel informational searches. When people are vaguely talking about the problem my product solves, having a product ad to my product seems too sudden. Like there needs to be more informing going on before they are presented with my product. Something I can only do currently with a standard search campaign.

 

[Steve]

So you have already maxed out the search volume for high search intent (buying) search phrases?
We only ever go broad in tiny markets where budget outstrips search volume (geographic campaigns, niche b2b).

Other wise start with skags based around obvious high intent keywords. Once that's maxed out add some modified broad match ad groups, constantly pulling successful keywords out into their own skag. Then move onto retargeting, and demographic based ads (display network, Facebook).

Playing wack-o-mole with negative keywords and broad match is only ever worth while if you are totally clueless about the market and your audience, the search volume is way smaller than your requirements, or you have a crazy client that specifically want brand awareness over sales (actually happens sometimes)

 

[Zach]

So you have already maxed out the search volume for high search intent (buying) search phrases?

No. I just went broad with it as a method to figure out what keywords are good and what are bad. Looks like I went too broad. But now I honed it in a bit and got some useful stuff.

My setup the last 30 days has been 13 different SKAG groups of tightly themed keywords.

From a cross-source analysis, here's how things have gone:

Last 30 days:

Source | # Sales | CPA
Facebook | 2,072 | $11.92
Google | 150 | $20.32

That $20 Google CPA is INCLUSIVE of a lot of probably lowerish intent keywords which I've just been trying to see what happens. I've got the data of the past 30 days of SKAG data to show ya.

Maybe try Wells Fargo, once they see your debt to income ratio and a history of your cashflow and balance they usually start sending you offers for lines of credit 300k-500k+, depending on what you having rolling in monthly. They have been good to us. Even though we haven't had to do that yet (take out credit to roll over), they are always there trying it seems. New banks are also hungry for new business, they are more competitive than most of the established banks out there. maybe try to hit them up also. If you are low risk to them you should have no problem getting a decent rated operating note from them to turn more sales.

hope this helps.

i ran out of time to add this to my reply, but the worst that will happen from you shopping around with other banks is that they will say no (you wont lose your birthday or anything haha), dont hurt to check around. sorry for the dbl post guys.

 

[Zach]

Below are the results of the 13 ad groups that been simmerin' the past 30 days:

Some great positives as well as big losers.

The way I see this data is things look particularly good for low CPAs + low search impression share ad groups. Like AG3/AG4. Still given the total amount of conversions on the table, this is just a low amount of data to make decisions on. I feel like I need to up the bids on everything pretty much, but lower them on the losing ad groups (5 of them in total).

For each ad group I tested 3 different ad texts & 2 different product URLs. I've got ad rotation set to manual so Google doesn't mess up the testing. Most the stuff still is at a point where it's not to a point of making a call on the best ad.

My 2 products:

1. Product A | 10 conversions | 1.7% CV | $.56 EPC
2. Product B | 23 conversions | 2.9% CV | $.75 EPC

I'd consider this data set on the light side, and product 2 got notably more than 50% of the clicks so not really a true A/B, but interesting to see. Ads were identical.

For most of you SEO is the thing.

These keywords I'm seeing that I get a "cheat sheet" into seeing have some pretty good CPAs......logically I'm thinking how to rank organically for these KWs.

However, the top competitors for all this stuff are the following:

healthline.com
medicalnewstoday.com
nih.gov
edu sites

Just feels like a really tough pack of sites to make a dent in. They are the types of sites that span thousands of topics with huge DAs with doctors writing and stuff. I just haven't even wanted to touch this from an SEO perspective because it intimidates the hell outta me. Google just has no incentive to give any of this traffic to our small, young brand. Thoughts?