How to Optimise Google Ads

  • What Bidding Strategy for Search Ads?
  • Should you use Portfolios?
  • How to Optimise Shopping Ads
  • Pro-tips to squeeze the most out of your ads
Image of co-founder

by Al Taylor
Updated June 28th 2021

Section 1:

Best Bidding Strategies for Search Ads

To optimise to a Target CAC it’s best to use Target CPA or Target ROAS

To optimise your Google Ads then it’s critical that you know your Target CAC.

Hitting your Target CAC will make you the most money (CM3) over the lifetime you have set (we suggest 2 years).

Find out more in the report: ‘Definitive Guide to Running an e-commerce Business’.

In that article we also discussed why focusing on Cost per Click (CPC) is not a good idea as it has no concept of the quality of the person clicking the ad.

Having lots of people click on your ad and not buy anything makes you no money but can cost you a lot even if the cost per click is really low.

In this article we’ll discuss how to set up Optimize Google Ads to get the most new customers that you can while hitting that Target CAC – including some pro-tips for squeezing out extra efficiency.

What bidding strategy to use for Search Google Ads (not Shopping)?

Given that we should be laser focused on Cost of Acquisition (CAC), then let’s use a bidding strategy that aligns with that.

Google has an option to optimise on cost per order (CPO) or sometimes called Cost per Action (CPA) [where an action could be an order, download, subscription or another event on your website].

You can also optimise on Target ROAS (Return on Ad Spend). [A ROAS target of 200% means the platform will aim to achieve £200 in revenue for every £100 of spend].

The good thing about Target ROAS is that it allows a higher cost per order on campaigns where the average order value (AOV) is higher (e.g. the AOV on a sports shoe campaign may be five times more than a sports sock campaign – with Target ROAS the platform will aim to spend five times as much for each sports shoes order).

Learn how to set up Target Cost per Order here: Google Ads: Learn about Target CPA bidding

Here is how you set up Target ROAS: Google Ads: Learn about Target ROAS bidding

Section 2:

Should you use Portfolios?

  • What is attribution
  • Why conversion path / customer journeys make perfect attribution impossible
  • GA default attribution: non-direct last click

You can also group your campaigns into a portfolio and set a target Cost per Order or Target ROAS at the portfolio level.

Then let Google optimise where to show ads to generate the most conversions (orders) or Revenue at that Target Cost per Order or Target ROAS respectively.

A portfolio allows Google to trade campaigns off against each other – it gives it the flexibility to choose which campaign to invest the next dollar into.

Let’s explain how this works with an example. There are three Campaigns (A, B and C) with a number of available conversions in each Campaign, but each conversion costs a different amount.

Screenshot showing example conversions available by campaign

What would happen if you set each campaign to individually have a CPO Target of £3?

You would get no orders from A; 6 orders from B and 4 orders from C ⇒ 10 total orders at a cost per order of (£2×6 + £1×3 + £7)/10 = £2.2 [a long way under your target CPO].

If these three campaigns were in a portfolio then Google can trade each campaign off each other to reach the target CPO.

In a portfolio it would get 6 orders from B, 4 from A and 4 from C ⇒ 14 total orders at a cost per order of (£5×4 + £2×6 + £1×3 + £7)/14 = £3

So under the portfolio scenario you would get 4 more conversions (40% more) at your Target CPO.

You can learn how to set-up a portfolio here: Google Ads: Learn about creating a portfolio

What campaigns to put in a portfolio?

We recommend putting all campaigns into a portfolio except for:

  • Brand Search campaigns (these are search campaigns triggered when someone is typing your company or proprietary brands).
  • Brand Awareness campaigns or campaigns that do not have a target cost per action.


How many portfolios to set-up?

You will want a different portfolio for each different goal type:

  • If you run a Lead Generation (which will need a Target CPA).
  • Purchase / Direct Marketing campaigns (probably best on Target ROAS).


If you are using Target ROAS then you can include campaigns in the portfolio that have different AOVs (Average Order Value) as long as the % gross margin (or ideally CM2%) is similar across them (though note that if the lifetime value is substantially different for particular products/categories then you want to keep separate and assign a different Target CAC to them).


  • If campaign A has an AOV of £100 with 10% CM2 margin it delivers £10 in absolute CM2.
  • If Campaign B has an AOV of £20 with 50% CM2 margin, it also delivers £10 in absolute CM2


It would not be a good idea to put both these campaigns in the same portfolio as if the Target ROAS was 400% then Google Ads would try to spend £25 on each order for campaign A giving a CM3 on that order of -£15 and Google would spend £5 on Campaign B delivering a CM3 on that order of £5.

You would want to put campaign B in a portfolio with a Target ROAS 5 times greater than the Target ROAS for campaign A.

If you have multiple product categories with very different life-time values, then you should set Target CACs for each category and each category should have its own portfolio of campaigns.

There is a trade off on the number of portfolios to use. Having more than one portfolio allows you to group campaigns that generate similar %CM2 (or even better similar LTV) together. Too many portfolios and there is not enough data for the ad platform to optimise (as well as fewer campaigns to trade off each other).

Keep an eye on Portfolios

Google doesn’t always run the campaigns within a portfolio as you would expect (i.e. works out efficienctly where to invest the next dollar of spend).

We have seen Google consistently deliver lower ROAS on particular campaigns even when other campaigns in the portfolio are not maxed out.

If you see this happening then you may want to split out the offending campaign and run outside of the portfolio or you can try using adjustments to the Target ROAS at the Ad Group Level as described below.

Adjust Portfolio Target ROAS at the Ad Group Level

There is another way to accommodate campaigns that need a different Target ROAS to the overall portfolio – or if you want to give certain campaigns within a portfolio a nudge.

You can keep or add a campaign to the portfolio but then set a Target ROAS at the Ad Group Level.

Screenshot showing Google Ads - adjust Target ROAS at Ad Group level within a Portfolio

You can also try using this if Google consistently drives a lower or higher ROAS on a particular campaign within a portfolio.

Section 3:

How to Optimise Shopping Ads

If you use Facebook Ads then you should be setting up UTM Parameters on every Campaign – the good news is that it is super simple

For simplicity we recommend Smart Shopping campaigns.

Learn how to set-up a Smart Shopping campaign

The default bid strategy for Smart Shopping is Maximise Conversion Value bidding.

It’s absolutely critical that you set a Target ROAS so that you optimise to a target return on your spend – without a Target ROAS you are optimising to spend your budget.

There is a period of learning before the system can optimise on Target ROAS so during this period you need to keep a close eye on your daily budgets.

Maximising Conversion Value means ‘spend all your budget’ while trying to get the most revenue.

It sounds good but it means Google spends all your budget even if there are no conversions to be had at a sensible cost per order.

Using the same example as earlier let’s see why using Maximise Conversion Value bidding without setting a Target ROAS (optimising to your budget) is a bad idea:

Screenshot showing example conversions available by campaign

If your Target Cost per Order is £2 and you set a daily budget of £30 then Google would acquire all the lowest cost conversions until it has spent your £30: 3 from C; 6 from B; 3 from A ⇒ 12 orders @ £2.5 an order. In this case you’ve gone over your target Cost per Order.

If you used Target ROAS to control spend then you would have got 1 from A @ £5 each; 6 from B @ £2 each and 3 from C @ £1 each. Total of 10 orders for £20 ⇒ Cost per Order of £2 and you would have avoided spending £10 on two extra orders (which at @£5 an order are 2.5 times more than your target).

This is not too bad as the most efficient overall spend (£20) is not too far off the total budget (£30).

But what if you had set the daily budget to £50.

Without a Target ROAS your average CPO becomes £3.33 (from A: 4 orders @ £5 each; from B: 6 orders @ £2 each; from C: 3 orders @ £1 each, 1 order for £7 and 1 orders for £8).

You’ve now spent an extra £30 to get 5 more orders ⇒ each incremental order costing £6 on average – three times your target CPO.

Google will keep acquiring the next ‘least expensive’ conversion (even if it is way above what you would like to spend) until it has spent your budget.

Without a Target ROAS you can also miss out when the going is good:

Let’s say you run a summer sale. The conversion rate on your site rockets as the discounts drive purchases.

This in turn means the cost per order with the same campaigns halves and there are more conversions to be had (If the conversion rate doubles then the cost per order halves – it costs the same to get the people to your site but twice as many people purchase).

Now Ad set A can drive 12 orders for £2.50 each; Ad set B can drive 18 orders at £1 each and Ad Set C can drive 15 orders (9 at £0.50; 3 at £3.50; 3 at £4).

With a budget of £30 and no Target ROAS you get 9 orders from C (@£0.5 each); 18 orders from B @ £1 each) and 3 orders from A (@ £2.50 each). With Cost per Order of £30/30 = £1

However with Target ROAS you get all conversions = 45 orders @ £1.67 per order ⇒ 50% more conversions. Without Target ROAS you have missed out on 15 conversions.

What happens if conversion drops on your site?

The summer sale has just ended and conversion rate drops – without Target ROAS you will spend all your budget but with Target ROAS you only get the conversions at an average of your Target ROAS and can save your marketing spend for a day when it is more efficient.

Without a Target ROAS, maximising conversion value optimising to a budget spend – but the chances that on a given day your budget spend matches to the optimal spend is low.

Setting up Smart Shopping Campaigns – should you focus on New Customer Acquisition?

When creating your Smart Shopping campaign there is the option to focus on either online sales to new and existing customers or on new customer acquisition.

Screenshot showing Google Smart Shopping option focus new customers

If you select New customer acquisition then you can assign an additional value that is ascribed to new customers on top of their actual conversion value.

As the tool is trying to maximise conversion value then it will disproportionately target conversion from new customers (the extent of which depends on the value you assign a new customer).

Google can identify a new customer either using its default conversion tracking detection (has the user visited the purchase page on your site in the last 560 days?)

Or you can upload a list of existing customers and then Google can identify those customers that are logged in to their Google Accounts.

We recommend that you select the ‘online sales to new and existing customers’ option and preferentially promote new customers by using audiences (see Pro-tip).

Section 4:

Google Ads Pro-Tips

  • How to use Audiences to optimise your campaigns
  • Use Google’s Attribution Tool to see if campaigns targeting new customers perform substantially better on a first click basis – if yes then you may want to push harder

Pro-Tip 1: Use Audiences to optimising campaigns between new and repeat customers

Our recommended bidding strategies of Target CPA and Target ROAS optimise based on conversions or conversion value.

These strategies on their own don’t differentiate between a new and repeat customer.

For most e-commerce businesses the cost of a new order is higher than the cost of a repeat order as existing customers will typically have a higher conversion rate [if a hundred existing customers click on an ad @ £1 a click and 10 convert then the cost per order is £10. If hundred non-customers click on an ad @ £1 a click and 5 convert then the cost per order is £20].

Take your most important campaigns and duplicate them.

Now assign an audience of ‘all converters’ to one campaign and exclude ‘all converters’ from the other campaign.

[For for Smart Shopping campaigns you can not exclude audiences but you can target an audience of non-converters].

You can learn about creating audiences here.

You have three options for identifying existing customers:

  1. You can use Google detection which identifies users that have visited the purchase page (with a maximum look back window of 560 days).
  2. Upload a customer list (Google then securely encrypts personal information and matches to user’s account ids).
  3. Conversion tracking using tags.

You can read more about how to provide signals for identification of existing customers here (scroll down to the section titled ‘How to distinguish new customers from existing customers’).

Note: You can add an audience to an existing campaign as an observation rather than targeting.

An observed audience does not change who the campaign is targeting but allows you to view the performance of that campaign by the observed audiences.

Outside of a portfolio you can also make bid adjustments on specific observed audiences differentiate the bidding on certain audiences.

The identification is not perfect

The only perfect identification of whether an order was new or repeat will come from your e-commerce platform.

The ‘All Conversions’ audience will take you some of the way and you can use mapflo to join up Google Ads data with your ecommerce platform data so you can accurately assign new and repeat orders to a specific campaign (on a last non-direct click basis) and use this data as a feedback loop to then change the Target ROAS on your Google Ads campaigns.

Read about mapflo’s solution to combine Google Ads and e-commerce platform data here

Pro-Tip 2: Use Google Analytics Attribution Tool to give insight on how new versus repeat campaigns perform on a first click versus last click basis

It’s not unusual at the account level to see a similar number of orders credited to Google Ads whether you use a first or last click attribution.

However at the campaign level there can be big differences – especially when comparing campaigns that are targeting new versus repeat customers.

It makes intuitive sense that new customers will be at the start of the purchase journey whereas repeat customers may be ready to buy.

The Google Analytics Attribution Tool allows you to view the performance of a campaign using different attribution models.

You can read more about the Attribution Tool here.

If you find that your new customer campaigns are given a lot more credit on a first click basis
versus last click (and likely that your repeat customer campaigns are the reverse) then you may want to push the spend on the new campaigns even harder (and pull back on the repeat spend campaigns).

Pro-Tip 3: Use Observation Audiences to identify high converting intent audiences

It’s worth adding Observational audiences to your campaigns. Observational audiences do not change who the campaign is targeting but allow you to see how well the campaign segmeted by those audiences.

If you are struggling to get a campaign to be viable (get conversions at a required cost per conversion) then you can see if any of the observed audiences are delivering at your target cost per conversion and try targeting the campaign on that audience.

Pro-tip 4: Use Target Audiences to control distribution of Shopping Ads

If you are selling products (such as T-shirts) where there could a hundred different products that could be shown for a particular search keyword (such as ‘Men’s T-shirt’) then you may find that Google ends up showing the same product again and again.

Even if you split out the products into individual campaigns the situation can still arise as all the individual campaigns are competing against each other to be shown for that broad search term.

You could divide your products across two or more Smart shopping campaigns that target different interest audiences allowing you to present a wider variety of product and learn more about which products are getting most interest.