The 70% overstatement hiding in your Google Ads new-customer reports

Here's a number that blew my mind when I finally dug into it.

Google told me we acquired 1,200 new customers in Q3. Our Salesforce data said 360 new companies. That's a 70% overstatement. Not a small discrepancy. A massive one.

And I'm not special. Every B2B account I've audited has this same gap. Sometimes it's 50%. Sometimes it's 80%. But it's always there, hidden in the default reporting.

Here's why, and how to fix it.

How Google counts "new customers"

Google's definition of "new customer" is account-based, not company-based.

When someone with email address john@acme.com clicks an ad and converts, Google flags that as a new account. Then later, when someone with email address jane@acme.com clicks an ad and converts, Google flags that as a second new account.

But that's the same company. Acme is not a new customer. Acme is an existing customer that now has two contacts in your system.

In B2B, this is massive. Because procurement doesn't work with one person. It works with a committee. An engineer finds you. Then they tell the procurement manager. Then the procurement manager gets their boss involved. Then the CEO gets looped in. Four people from the same company, four different email addresses, four "new customers" in Google's reporting.

You're selling a $100,000 system. One company buys it. Four decision makers are involved. Google counts that as four new customer acquisitions and a $400,000 revenue impact (if you're tracking revenue based on the new customer count).

Your Salesforce says you acquired one new company and one $100K deal.

Both are true. But Google's number is inflated by 4x.

The math on this

Let's use realistic numbers for an industrial manufacturing account.

Your actual new companies acquired: 60 per quarter Average decision makers per deal: 2.5 to 3.5 depending on deal complexity Google's "new customer" count: ~180-200

Google reports it as 70% more acquisition than actually happened.

Now let's say your average deal value is $50,000. And you're tracking "new customer to direct" revenue in Google Ads.

Google sees: 180 new customers times $50K = $9M new customer revenue Salesforce says: 60 companies times $50K = $3M new company revenue

Your blended ROAS looks 3x better than it actually is. Your CPA looks 3x better. Your entire reporting foundation is built on inflated numbers.

And if you're using that data to optimize bids, you're bidding like you have a 3x better performing account than you actually do. Which means you're underbidding relative to the reality of your conversion rate.

Why this matters for bidding

If Google is telling you that your account has a better cost per new customer than it actually does, you're probably underbidding.

The algorithm sees your "efficient" new customer rate and thinks "cool, this account is performing great, we can maintain these bids or even reduce them." But the actual new company rate is 3x worse. So you should be bidding more aggressively to compensate.

Instead, you're bidding conservatively based on inflated data.

Over a quarter, that means missed opportunities. Budget sitting on the table. Deals you could have won going to competitors because you underbid.

How to fix it

You need to deduplicate by company domain.

Pull your Google Ads conversion data (new customer to direct, or whatever your SQL conversion is). Pull your Salesforce new company data. Join them on domain.

Here's the BigQuery logic:

WITH gads_conversions AS (
  SELECT
    SUBSTR(email, STRPOS(email, '@') + 1) as domain,
    COUNT(*) as gads_new_customers
  FROM conversions
  WHERE conversion_type = 'new_customer'
  GROUP BY domain
),
sfdc_companies AS (
  SELECT
    LOWER(SUBSTR(website, 
      CASE 
        WHEN website LIKE 'https://%' THEN 9
        WHEN website LIKE 'http://%' THEN 8
        ELSE 1
      END)) as domain,
    COUNT(DISTINCT id) as actual_new_companies
  FROM opportunities
  WHERE close_date >= '2023-01-01'
    AND is_closed_won = true
    AND is_new_business = true
  GROUP BY domain
)
SELECT
  g.domain,
  g.gads_new_customers,
  s.actual_new_companies,
  ROUND(g.gads_new_customers / NULLIF(s.actual_new_companies, 0), 2) as overstatement_multiplier
FROM gads_conversions g
LEFT JOIN sfdc_companies s ON g.domain = s.domain
ORDER BY overstatement_multiplier DESC

Run this and you'll see your actual overstatement multiplier per domain.

Some domains will be 1x (Google's count matches reality). Some will be 5x or higher (companies with big procurement teams).

The real fix: domain-level new customer tracking

Once you see the multiplier, you can adjust.

Instead of using Google's "new customer to direct" metric as gospel, create a custom conversion that represents "new domain to direct." Track the first conversion from each domain per month, not the first conversion per email.

This is more work upfront (you need to deduplicate in your tracking tags). But once it's set up, your reporting becomes real.

Then you can actually optimize against your real conversion rate, not Google's inflated version.

Or, if you can't modify your tracking, just apply the multiplier retroactively. If your overstatement is 70%, divide your "new customer" numbers by 1.7 before you make optimization decisions.

It's not perfect. But it beats making decisions on data that's off by a factor of three.

The pattern

This is one example of a broader issue in B2B Google Ads reporting: the metrics are designed for consumer e-commerce, where one person is one customer.

In B2B, one company is one customer, but multiple people are involved. Google doesn't understand that distinction. So its default metrics are inflated.

Until Google fixes the reporting (which they won't, because inflated metrics make their platform look better), you need to fix it yourself.

The deduplication logic above works for any account where multiple stakeholders are involved in procurement.

Which is basically all of B2B.

So if you haven't audited your actual new company rate against Google's new customer count, do it. The gap will probably surprise you.

And if you're making optimization decisions based on Google's inflated numbers, you're leaving money on the table.