Criteria Corp is an HR tech SaaS startup that helps companies assess, interview, and develop talent both pre- and post-hire.

In 2022, Inc. named Criteria one of the 5,000 fastest-growing private companies for the 8th year in a row.

Criteria, which was founded in 2006, began with a narrow focus: as a pre-employment assessment provider for small- and medium-sized businesses.

By 2023, the company had an ever-expanding product portfolio and a growing number of employees, and the natural question as a business was how to move up-market in order to fuel the next phase of growth.

It just so happened that I, as Senior Digital Marketing Manager, was on the Criteria demand generation team and was tasked with figuring out how to bring additional mid-market and enterprise leads in the door!

While the obvious answer, of course, is account-based marketing (ABM), this is a long-term strategy that requires significant internal alignment between sales and marketing on execution and can take quite a while to bear fruit.

To be clear: ABM was certainly part of the mix in terms of Criteria's go-to-market in targeting potential/additional mid-market and enterprise revenue, but those efforts are beyond the scope of this case study.

Luckily, when the request was made from on high to bring in additional mid-market and enterprise leads (and by extension pipeline and revenue), we had already spent over a year ensuring the necessary data was in place.

In many ways, UTMs (Urchin Tracking Modules — Google purchased Urchin in 2005) form the core of web analytics. UTM Medium and UTM Source are particularly valuable, especially when combined.

Even casual users over the years of Google Analytics (which Urchin eventually became) have no doubt noticed that UTMs provide essential information on where traffic comes from.

More advanced users and analytics managers build custom UTM tracking links that can unlock more campaign-specific custom reporting that can be useful especially for conversion attribution.

It takes considerable expertise, however, to capture UTM values and store them in a company's Salesforce CRM — and this is what took us many months to get completely right.

To make a long story short, we:

Each Salesforce Lead contained full UTM information and much more. Our Salesforce admin was kind enough to make sure these fields passed all the way to the Salesforce Opportunity and Order objects as well.

Once we had this (and it took a little while to iron out the kinks, since inevitably problems arise in the data), it was like shooting fish in a barrel.

By running reports and creating dashboards in Salesforce, we could see which traffic sources were leading to pipeline and revenue, and which were not. (It was like suddenly turning a light on in a dark room.)

Furthermore, we quickly realized enough SMB leads were coming from free/organic traffic sources to cover our target (i.e., the number of leads we owed the sales team each month).

Of course, this is just another way of stating the obvious: there was no shortage of SMB leads.

Nevertheless, this was an encouraging sign. By having this kind of data in hand, we no longer had to assume anything about where to put our focus. Specifically, we wanted to take our Google Ads account to the next level.

It is important to reiterate that everything you are about to read about in this section would have been completely impossible if not for the work done as described in the previous section.

For instance, I noted that UTMs were not the only fields captured and sent to Pardot (and ultimately Salesforce).

When configuring a Google Ads account, it is critical to use a tracking template that automatically adds important parameters to the query string after an ad is clicked.

It is also important to note, however, that the most crucial of these parameters — the GCLID (Google Click Identifier) — is automatically added whether or not a tracking template is present.

Each time someone clicks on a Google Ad, a unique string of randomized numbers and letters is generated and added to the end of the landing page URL like this: website.com?gclid=123456abcdef

Remember those first-party cookies I mentioned earlier? Well, one of them was configured to automatically store the value of the GCLID and pass it into forms on the Criteria website via hidden fields.

This may not seem like a huge development, but storing the GCLID in our Salesforce CRM opened an entire world of performance to us.

We could see which leads (which, by nature of filling out a Free Trial or Pricing Request form on the website, were MQLs by default), converted to Sales Accepted Lead (SAL) or Sales Qualified Lead (SQL).

By generating a simple Salesforce report, I could easily track in real time which leads converted to SAL (and then SQL). I could see that lead's GCLID and company size (SMB, MID, ENT).

Then, through a process that was painfully manual, I created and continually updated a spreadsheet with some critically important columns: Google Click ID, Conversion Name, Conversion Time, Conversion Value, Currency, etc.

Conversion value is designed to do exactly what it says: convey the value of a conversion. In an eCommerce context in particular, this is pretty straightforward.

For a B2B SaaS company like Criteria Corp, it's a little more complicated. Ideally, ARR (annually recurring revenue) would be what we would use and that would be that — we wouldn't need to bother with SAL, SQL, etc.

But alas, there are technical limitations to the GCLID, namely that it can only be used for 90 days — if you upload a conversion to Google and the click happened longer than 90 days ago, the conversion won't count.

For us, this was truly unfortunate: most mid-market and virtually all enterprise deals had sales cycles that were longer than 90 days. We weren't able to import closed won revenue for enterprise and optimize for that, sadly.

There was also a question of conversion volume. We were looking to get results as quickly as possible, and since SAL is the next step down the funnel after MQL, this meant starting there.

In our Google Ads account, we set up a conversion for SALs and selected it as the account's primary conversion. We also set up a conversion for SQLs, but left it as a secondary conversion (i.e., one not used for optimization).

At first, we wanted as many SALs as possible to ensure we were getting enough. (To clarify: before we began importing conversions containing GCLID information and funnel status, we were only able to optimize for MQLs).

This meant selecting "Maximize Conversions" as our bid strategy for the first couple of months. As expected, the number of SALs from Google Ads began to increase.

However, the objective was not more SALs overall (although this increase was not unappreciated), it was more mid-market and enterprise SALs.

By maximizing conversions, we were getting more SALs, yes, but the lead mix in terms of company size still was the same.

This is where we got the most creative. When we imported our conversions, we assigned a conversion value based on company size. To keep things simple, we used the average ARR of each segment as the conversion value.

It is important to note that Google takes user privacy very seriously and has to remain in compliance with laws such as GDPR in the EU, CCPA in California, and particularly COPPA, which protects children.

Since Google must remain in compliance, that means that we as users of Google's products must be exceedingly careful not violate privacy laws when importing first-party data or risk getting our Google Ads account banned.

The GCLID is what allows for a user's data to remain anonymous. Of course, when a user fills out a form on the website, we can then see their personally identifiable information (PII) in our marketing automation platform and CRM.

Feeding PII back into Google without that user's consent would be a violation of data privacy laws, but using the GCLID as a proxy is completely fair game. Google is then able to take the anonymized data and build a model with it.

It's important to remember that Google does not know which of its users work at enterprise or mid-market companies. You cannot just ask Google to only show your ads to users who work at companies of certain sizes.

Can you do so with LinkedIn Ads? Yes, but LinkedIn Ads, like all social media ads, interrupt the user. Search ads are based on intent — get in front of a user that is actively searching for what you are selling and you'll create demand.

Again, it's important to keep in mind that we were working with a blank slate in the beginning. Google's algorithm did not intrinsically understand what we were setting out to do, and never did — it just did what we told it to do.

In other words, this was an extremely customized solution to a complex problem. It also took a long time to execute, with me importing the conversion data via a spreadsheet manually for an entire year.

As we fed more and more conversions into Google, we pushed Google's AI (machine learning) to find more users like the ones that had converted for larger "amounts" (i.e., MID and ENT) and show our ads to them.

This also meant that we continuously uploaded SMB conversions as well as a way of telling Google the kind of users we did not want. (Target ROAS requires this balance, since you're optimizing for conversion value.)

Slowly but surely, the mix shift for SALs began to move dramatically — and we were asked to trim our spending month after month as our quotas were more easily hit. Within 6 months, the entire sales team was reorganized.

Whereas before we had our best Sales Executives handle what enterprise leads came in, those same SEs would inevitably have to dip into MID and even SMB leads because there just weren't enough ENT leads.

Suddenly, things were different: now there were enough enterprise SALs for those SEs to focus exclusively on new enterprise revenue and not have to worry about adjusting their approach for MID and SMB.

Similarly, a new group of SEs were tasked with selling exclusively to mid-market leads now that there were enough mid-market SALs. This reorg enabled greater alignment with SEs' natural strengths and selling styles.

After a full year of importing GCLID and other first-party data back into Google, we were sitting pretty.

A year earlier we were lucky to crack 30 ENT SALs in a month. Now we were consistently doing 90 or even 100 — while spending 30% less than we were a year earlier.

What's more, the business had already (in September) blown past its yearly target for new enterprise ARR.

When the target had been set in January, none of this future impact (including the sales reorg on April 1) was conceivable.

This is where this particular story ends — although the work continued, the tactics didn't change, which is why I'm stopping the story here.

However, I will tie up a couple of loose threads you may be wondering about: