Optmyzr Blog

3 Ways to Improve Smart Bidding Performance on Google Ads

Jul 08, 2021
Smart Bidding

Frederick Vallaeys

Co-Founder & CEO



Bid automation is one of the biggest potential time-savers in PPC. Machines are simply faster and more reliable than humans at constantly updating bids based on hundreds of factors influencing conversion rate and conversion value per click.

But what if you’re not seeing the results you were hoping for? It could be that the bid automation isn’t the problem but that you didn’t give it all the information needed to make smart decisions.

For example, when you reported a conversion and told Google that the customer spent $100 in your online store, could you have shared better data like how much was returned and how much profit you made from the items the customer kept?

The point is, you can never go hands-off with any kind of automation. Instead, what you should do is limit the time you are involved.

Optimizing in this way by manipulating conversion values is a newer territory for many advertisers. The value data already flows through conversion tracking, and using that as an optimization lever has traditionally not been top of mind for many advertisers.

Optimizing conversion value should be a priority in a more automated PPC world, and there are 3 options Google offers advertisers. In addition to working with your own web development team, these are, of course, to send the value you want to Google through the conversion pixel in real-time.

So let’s take a look at the 3 options to share better data with Google’s machine learning systems about the value your PPC ads are delivering

Option 1: Offline Conversion Import (OCI)

Offline conversion tracking (or Offline Conversion Tracking) is the original way to tweak conversion data after something has already happened on your site.

It was primarily built for lead gen advertisers. But it’s surprisingly useful for eCommerce advertisers too. The idea is to capture a click identifier (called the GCLID or Google Click ID) and hold on to that in your own system until you’re ready to report that something desirable happened from that click.

At that time, you can send Google the GCLID, the conversion, and a value, and they are then able to connect that with all the attributes of the click.

That means the Google machine learning system can then start to figure out if any patterns correlate with users who tend to take the desired actions.

So, for example, if they notice many people from Germany convert and return high values for lower click costs (hence delivering high ROAS), then the automations may decide to try to get more clicks from Germany.

With OCI, you can’t restate values, but advertisers can create new conversion types and add them to ‘conversions’ where Google’s smart bidding algorithms will use them.

In lead gen, that could mean creating offline conversions for ‘qualified prospects’ after a salesperson speaks with the prospect on the phone and a ‘closed deal’ when that prospect eventually purchases from their sales rep.

Retailers can also benefit from this capability by nudging Google’s systems towards what they believe to be better customers.

For example, creating a secondary conversion based on a scoring model predicts a customer’s future purchase behavior with the retailer.

OCI is particularly useful in this case because it may not be possible to calculate that lifetime value component in real-time while the user goes through the checkout process. When it is calculated, later on, it can then still be communicated to Google.

Option 2: Conversion Value Adjustments

This second option of adjusting conversions is generally less known but it’s actually an easier to implement variation of OCI.

Rather than capturing and storing the GCLID, advertisers can add a transaction ID to their existing conversion tracking code from Google at the time of the initial conversion. Then they have 55 days to restate the value of that conversion.

Here’s an example of how that could work. The advertiser sells $100 worth of goods, but $40 of goods are returned by the consumer before the return window closes. The initially reported value of $100 can now be restated by telling Google the original transaction ID of that sale and the new value, in this case, $60 ($100 - $40 which was returned).

By doing this, Google’s automations may come to realize that people in Germany return more stuff than those in the UK and appropriately tweak its decision-making process to account for that factor.

While this system was designed to help retailers, lead gen companies can also benefit from it.

For example, they could retract a lead from someone who filled out a form but then never took advantage of the free trial offered by the sales team.

Option 3: Conversion Value Rules (beta)

The last option to teach the machine about value is to deploy conversion value rules which are currently in beta testing from Google.

This is a less precise methodology because it doesn’t allow advertisers to tweak individual conversions. It does however allow advertisers to tweak conversion values based on some high-level attributes that they may have already correlated with desirable behaviors.

As in the previous example of buyers in Germany and the UK, they could boost their reported conversion values for users from the UK because they do fewer returns. Other attributes around which these rules can be built are device types and audiences.


Smart bidding can seem like a better alternative to letting Google do the heavy lifting for you. It’s perfect for automating the more tedious parts of PPC management so that you can spend time on strategy.

But smart bidding is just one part of the bigger puzzle. You are still required to perform keyword research, write compelling ad text, optimize your ad text, set up campaigns, etc., to make sure you are meeting your business goals.