We Tested Three Ways to Enrich Inbound Leads. Here’s What We Found.

Up-to-date company data matters when you’re scoring and routing inbound leads. But which enrichment approach gives you the information you need, and how much can you trust it?

To find out, we ran our own four-week experiment at With Scale. As 300 companies entered our workflow through real inbound leads, we ran three enrichment approaches alongside each other:

• HubSpot’s out-of-the-box enrichment, which has no additional cost and doesn’t use HubSpot credits

• Clay’s out-of-the-box enrichment, which uses one Clay credit per company for full enrichment

• Clay’s advanced waterfall enrichment, which uses up to 6.4 Clay credits per company, per enriched data point (for the properties we were enriching)

We compared how quickly the three approaches returned results and how much their employee count and annual revenue figures varied across 300 companies.

So, what did we learn?! Here are our key takeaways:

• Speed is unlikely to be the deciding factor. For the most part, Clay and HubSpot both enrich within the minute, which is great for lead processing times and maintaining speed to lead.

  Clay’s results had the same minute-level timestamp as HubSpot’s in 83% of cases. In the remaining cases, they appeared in the following minute.

• Employee counts can vary widely between providers. For half the companies we enriched, the data variation between the systems was ‘medium’ – i.e. on one platform it may say the employee size is 25, and on another 52 (which is fine but not fantastic). In some other cases, estimates differed by as much as 100 times!

• When we spot-checked against LinkedIn, we found Clay’s enrichment was generally closer to LinkedIn’s.

  Keep in mind that, in most cases, we use ranges for lead scoring. As long as the tool you end up using falls within that range, it is probably accurate enough.

• Treat annual revenue as an estimate. Providers returned different figures and sometimes no figure at all. Clay’s basic enrichment returns revenue ranges, which we’d favour over specific estimates when assessing ICP fit for lead scoring.

What does this mean for you in practice?

• If you have Clay and care about employee size and annual revenue, we’d use it for inbound enrichment.

• If you don’t have Clay, don’t panic! HubSpot enrichment may still work, depending on how broad your employee-size brackets are. If HubSpot returns 55 employees but the actual count is 140, both fall within a 51–150 bracket.

• If your brackets are narrower and employee size is central to your ICP, we’d consider Clay.

This experiment focused on employee size and annual revenue, comparing speed and variation between providers. For more custom enrichment fields, Clay can be super helpful!

If you have monthly HubSpot credits, try its AI properties and see how they compare on more complex enrichment prompts.

Inbound lead enrichment key learnings breakdown:

1. Speed: Clay and HubSpot returned results within minutes

If speed to lead is on your mind, here’s the good news…in this test, all three approaches were quick.

HubSpot’s native enrichment populates as soon as a record enters the workflow, as long as it finds the record in its database. It’s pulling from data that’s already available, rather than running a new search.

The two Clay approaches returned results at the same time. In 83% of cases, their results had the same minute-level timestamp as HubSpot’s. In the remaining cases, Clay’s timestamp was, at most, one minute later. That could simply mean HubSpot’s enrichment happened towards the end of a minute.

Honestly, Clay was faster than we expected! Based on these results, the small difference in recorded times isn’t enough to choose between the options.

2. Employee counts: check how much the providers disagree

Next, we compared the employee counts returned for each company. We ordered the companies by the gap between the highest and lowest counts, measured as a ratio, to see how much the three enrichment options varied.

For the first half of that list, the providers’ employee counts were closer together. For example, if the lowest estimate were 100 employees, the highest would have been below 200. After that, the differences grew sharply, reaching as much as 100 times for some companies.

That sounds dramatic, but ratios need a bit of context. If one provider says a small company has 2 employees and another says 20, that’s a 10x difference. Meanwhile, a gap between 14,000 and 24,000 employees is 10,000 people, but it shows up as less than 2x. So the ratio alone doesn’t tell you the whole story.

The graphs below show that HubSpot returned the highest employee count most often, while Clay’s waterfall enrichment most often returned the lowest. It would be tempting to call Clay’s basic enrichment the most accurate, but these results don’t establish which count is right.

These results show where the providers disagree; they don’t, on their own, tell us which count is accurate.

That said…when we spot-checked against LinkedIn, Clay did win out.

​That pattern still holds when we look only at companies where the highest employee estimate was at least three times the lowest. HubSpot most often returned the highest count, and Clay’s waterfall enrichment most often returned the lowest.

So, what might be behind those bigger gaps? We have a few hypotheses:

• Large companies with offices worldwide may be harder to count consistently.

• Higher education institutions may have LinkedIn profiles that include people who attended as students, rather than worked there.

• Very small companies, or those with little LinkedIn presence, may be harder for providers to find and identify.

• Company names and domains that don’t obviously match could make it harder to connect the right records.

• An incorrect company name in the CRM could also skew the result, even when the domain is right. For example, Travel Supermarket entered HubSpot as “Compare Travel Deals”.

These are possible explanations, not confirmed causes.

For around half of the companies with large differences between providers, the employee counts were already high. In several we checked, LinkedIn showed “10k+ employees”, while the enrichment sources returned counts ranging from 14,000 to 24,000. Depending on how a team sets its segments and routing rules, a difference at that scale may have little practical effect.

But that’s not true across the board. We've also seen plenty of smaller companies with large differences between estimates, so it’s worth looking more closely before relying on the number for segmentation or lead routing.

3. Annual revenue: use estimates with caution

Annual revenue is trickier to assess than employee count. For non-publicly traded companies with reported figures, providers estimate it. They may use headcount and an industry benchmark for revenue per employee, then adjust for factors such as funding stage or web traffic where available.

And as we’ve already seen, headcount figures can be out of date or vary widely between providers. Any error there can carry through to the revenue estimate.

Revenue data also had more gaps. Of the 300 companies in our test, HubSpot returned no annual revenue figure 37% of the time, Clay’s basic enrichment 10% of the time, and Clay’s waterfall enrichment 12% of the time. All three returned a result for 176 companies.

For the comparison graph below, we focused on HubSpot and Clay’s waterfall enrichment. Both return a specific number. Clay’s basic enrichment returns a range, which is harder to compare numerically, so we used it as a reference alongside the two specific estimates.

The differences between those specific estimates could be substantial. Even once we set aside the companies with the largest gaps, the higher estimate reached seven times the lower one in the remaining group. Only a handful had estimates less than twice as far apart. That makes revenue a figure to treat with a big pinch of salt when it comes from an enrichment tool.

We couldn’t verify which estimate was right for most companies, or identify a clear type of company where estimates were more dependable.

The least variation appeared among large publicly traded companies: Starbucks, Cencora and Tripadvisor had identical revenue figures from HubSpot and Clay’s waterfall enrichment. That suggests a useful check for a GTM enrichment workflow: flag publicly traded companies so your team can place greater confidence in their revenue figures.

For lead scoring, we’d favour Clay’s basic revenue ranges over treating a precise estimate as fact. Its tiers get broader as revenue increases, such as 1M–5M, 5M–10M, 10M–25M, 25M–75M and 75M–200M. If revenue helps determine ICP fit, broad bands may be more useful than a specific number that the providers can’t agree on.

How we did it – the nitty-gritty

The process

A HubSpot workflow started when a company-level field was automatically set to show that an inbound lead had arrived. HubSpot’s out-of-the-box enrichment then ran, while a webhook triggered both enrichment approaches in Clay.

We set one timestamp when HubSpot enrichment succeeded and another when Clay enrichment succeeded. We also created six properties to store the employee count and annual revenue returned by each of the three approaches.

How we monitored the results

We let companies enter the process as real inbound leads arrived. Over four weeks, 300 companies went through it.

Before analysing the results, we used a HubSpot view to monitor general trends and identify cases where one or more approaches failed, as shown by a missing timestamp.

A note on the experiment

This experiment measured how much the providers’ results differed. It did not necessarily establish which provider was most accurate against a verified figure.

For most companies, there was no independently verifiable “correct” employee count (even when we checked manually on LinkedIn) or annual revenue figure to check against. These figures are largely self-reported, estimated or modelled by the providers.

Public companies with audited financials are an exception, and we’ve mentioned relevant spot checks above, but this was not a systematic verification exercise. The findings show how much enrichment providers can disagree, and why that matters when using their data for lead scoring and routing.