Using Google Adwords ROAS bidding

Using Google Adwords ROAS bidding

At we like Adwords' ROAS (Return on Ad Spend) bidding method as we believe it provides the best way to obtain the highest return on investment on Adwords spend.

Here is how we configure our Adwords campaigns to maximise return on spend.


First we need to decide how to track conversions for our Adwords campaigns. The conversion funnel for our website is as follows: virtual account opening -> real account opening -> deposit funds -> trading.

However there will be a drop-off at every step of the funnel: only a percentage of virtual accounts will upgrade to a real account, only a percentage of real accounts will make a deposit, and deposits will take time to be traded fully.

Choosing the right conversion metric hence requires a number of judgment calls. If we track only new virtual accounts and/or new real accounts, we run the risk that Adwords will optimise to users who open accounts but never deposit. If we track only deposits or trading activity, we run the risk that the number of conversions will be too low for Adwords to properly optimise the campaigns.

Hence, we need a balance between these two extremes. We therefore decided to track new virtual accounts, new real accounts, and deposits - assigning to each conversion a dollar value determining by making an educated guess about the expected value of such a conversion to the business.

  • For deposits, we assign a value proportional to the deposit value.
  • For new real accounts, we assign a value which is a percentage of the CLV (client lifetime value) for clients based in each particular country (this is because CLVs vary considerably by country).
  • For new virtual accounts, we assign a percentage of the value assigned to new real accounts.

Overall, we calibrate the figures so that over time aggregate conversion values from new accounts are roughly equivalent (or a slightly bit higher) to aggregate conversion values from deposits. The ratio between these aggregates needs to be set by looking at the number of conversions of each type and making a judgment call accordingly.


Ultimately, our goal is to maximize ROI (return on investment), i.e. the ratio between aggregate conversion values and amounts spent.

The ROI we can hope to achieve is going to decrease as a function of amount spent. Indeed, if we spend very little, then Adwords will be able to pick the 'low hanging fruit' and achieve a high ROI. If we spend a lot, Adwords will be struggling to bid profitably and ROI will drop.

Therefore the goal will be to maximize ROI for a given level of spend.

Bidding strategies

Adwords provides a number of different bidding strategies: CPC, CPM, CPA, maximize clicks, In theory, ROAS. ROAS (return on ad spend) is exactly what we are looking for, as it seeks to maximize the ROI of the campaigns. According to Google's description, ROAS "adjusts bids using real-time signals like device, browser, location, time of day, etc."

However experienced Adwords professionals can be wary of adopting automated ROAS bidding because it strips them of all control over the campaigns. Of the very numerous controls that Adwords provides over campaigns, only 3 remain with ROAS: setting a maximum cost per click (which Google does not recommend for ROAS), setting the device bid adjustment to -100% for mobile devices, and setting the Target ROAS. Everything else is on autopilot!

Adwords professionals don't like giving up so much control to Google because they want to be adjusting bids themselves according to location, time of day or day of week, device, browser, ad group, keyword etc. They also do not trust Google's ROAS strategy to be designed in favour of Google's clients. Google says its motto is "don't be evil" but clearly Google is in a position of conflict of interest as regards bid amounts: how do we know that ROAS will be bidding in the optimal manner in favour of the client?

We know that because it is actually in Google's interest for its clients to achieve the highest possible ROI on their campaigns. Clients have ample other opportunities to advertise online: many other ad platforms exist (e.g. Facebook, Twitter, and many more) and clients are going to be allocating their spend to the platforms that provide the highest ROI. It is in Google's interest that its clients achieve the highest possible ROI on their campaigns.

But then how do we know that Google's automated ROAS system will outperform what an experienced Adwords professional can achieve with the more traditional CPC/CPM bidding methods? The answer is because there are too many moving parts for a human being to beat an automated algorithm. Bids can be adjusted by location, time of day, day of week, device, browser, operating system, ad group, keyword, whether the client is on a remarketing list or not, demographic information and browsing behaviour and history that Google may have - amongst many other factors that Google has at its disposal. ROAS intersects all these variables in N-dimensional space in a way that no human being can ever hope to achieve.

ROAS is available in two types: either as a standard strategy for a single campaign, or a portfolio strategy across multiple campaigns. We recommend using the portfolio strategy (i.e. ROAS set up via the Shared Library) so as to maximise the amount of data that ROAS can use to optimise the campaigns.

Managing the ROAS strategy

Once the ROAS strategy (from the Shared Library) has been assigned to campaigns, the only available parameters that can be tweaked to manage the bidding are:

  • The bidding adjustment for mobile devices, which can be only set to -100% to exclude mobile devices altogether.
  • The maximum per-bid amounts (which Google recommends not to set for ROAS campaigns).
  • The target ROAS value. The higher the value, the lower the aggregate spend will be. Importantly, if you see any campaigns that are showing "Limited by budget" then that means that you need to increase the Target ROAS.
  • The conversion amounts assigned to new virtual accounts and new real accounts, calibrated by user location. Importantly, these will need adjusting to achieve the desired ratio of new account conversion values -to- deposit amounts conversion values, to adequately front-load and back-load the client conversion funnel, depending on the number of conversions of each type.

Furthermore, if there are several campaigns using the same shared ROAS strategy, it is important to monitor the ROI of each campaign, and to remove the campaigns that are bringing down the average overall ROI.