B2B SaaS Case Study

How GrowthSpree Helped DataHub Triple Its High-Intent Leads

A paid search transformation from cheap content-download conversions to sales-rated, pipeline-generating leads, with high-intent lead share more than tripled.

1.6% → 5.3%High-Intent Lead Share
−50%Cost per High-Quality Lead
+79%Deals from New Build

DataHub

Metadata platform for data and AI

Founded
2021
Headquarters
Palo Alto, California
Company size
51–200 employees
Sector
Data management SaaS

Case study summary

GrowthSpree rebuilt DataHub's Google Ads around sales-rated lead quality, lifting high-intent lead share from 1.6% to 5.3%, halving cost per high-quality lead and growing deals from the rebuilt campaigns by 79%.

  • GrowthSpree more than tripled the share of DataHub's paid-search leads rated High by sales, from 1.6% to 5.3%.
  • Cost per high-quality lead fell 50%, while overall cost per lead stayed roughly flat.
  • Display and Demand Gen, which took 42% of spend and produced no High-rated leads, were retired to 0%.
  • Deals from restructured campaigns grew about 79%, and their share of attributed pipeline value rose from 42% to 67%.
  • Performance Max grew from 5% to 70% of conversions on 22% of spend, while themed Search delivered the highest-quality leads.
Where they started

A low cost per lead built on content downloads that never reached a sales conversation.

What we changed

Display and Demand Gen retired, broad match removed, campaigns split by product theme and region.

Where they are now

High-intent leads 1.6% → 5.3%, cost per quality lead halved, and 79% more deals.

Who is DataHub?

DataHub is a metadata platform that helps data teams discover, govern, trace and monitor their data. Its paid search program has to reach technical and data buyers across several product themes and regions, and turn them into sales-rated pipeline.

Client profile
CategoryMetadata platform / data catalog (B2B SaaS)
Product themes advertisedData governance, data lineage, data observability, metadata and AI, plus brand, competitor and financial services campaigns
MarketsUnited States, European Union and global
ChannelsGoogle Search and Performance Max (after the rebuild)
Definition of successLeads rated High by sales, attributed deals and pipeline value in the CRM

Why a low cost per lead was misleading

DataHub's account looked efficient because its cost per lead was low. But most of those leads were gated content downloads from Display and Demand Gen that entered the CRM and stopped there. None were rated High by sales.

The old metric

Cost per lead

Cheap, and inflated by content downloads. It counted form fills, and few of those were buyers. Reported CPL was low because the account was paying for the wrong thing.

The real metric

Cost per qualified lead

Higher on paper, but every lead behind it is a sales-rated, pipeline-generating conversation. The value of each lead rose with it.

Where the account started

21 live campaigns across six channel types, structurally unable to show which spend produced revenue.

Four structural problems

Budget in the weakest channels. Display and Demand Gen took 42% of spend to produce "conversions" that were content downloads, not buyers.
Broad match running unchecked. 71 broad keywords consumed 31% of Search spend at more than double the account's cost per lead.
Campaigns blended geo, theme and funnel. One budget bundled several products and regions, so no theme could be measured or defunded on its own.
Ads pointed at 26 pages, including docs. Developer documentation and expired webinar pages absorbed paid clicks, and half the pages produced zero conversions.

The six changes we made

Retire the waste, impose discipline and structure, fix where clicks land, then let Performance Max and Search each do their own job.

1

Retire Demand Gen

On platform metrics, Demand Gen looked like a top performer. The CRM said otherwise: of the leads sales scored, about 72% were rated Low, the rest Medium, and none High. It never appeared against a closed deal, so it was wound down in stages and fully retired.

2

Retire Display

Display's "conversions" were gated content downloads that entered the CRM and stopped. The deals it was credited with carried no pipeline value. It was switched off with the rest of the legacy build.

42% → 0%Display + Demand Gen spend
0High-rated leads from Demand Gen
NonePipeline from Display
3

Enforce match-type discipline

Broad match was removed. Search ran on phrase and exact only, split by intent, with brand isolated on exact match. One broad keyword alone had been taking a third of Search budget.

31% → 0%Broad-match share of Search spend
71 → 0Broad keywords serving
VerifiedIn-platform
4

Bifurcate campaigns into a theme × region grid

Blended campaigns were split so each product theme (data governance, data lineage, data observability, metadata and AI, competitor, financial services, brand) in each region (US, EU, global) got its own budget, keywords and landing page. Performance became measurable, and one theme could be funded or cut without touching the rest.

21 → 16Live campaigns
6 → 2Channel types
1 : 1Theme to region per campaign
5

Build landing pages for relevance

Paid clicks moved off developer documentation, expired webinars and blog posts onto product pages matched to each campaign theme. Pages were consolidated onto ones the team controls and tests.

36% → 79%Spend on product pages
12 → 5Zero-conversion pages
4.8% → 1.2%Wasted landing-page spend
6

Use Performance Max for volume

Performance Max campaigns were scaled, including an EU product campaign, and tuned to fill the funnel efficiently while themed Search supplied the qualified conversations.

5% → 70%Share of conversions
22%Share of spend
−61%PMax cost per conversion
How the rebuild was sequenced
StageWhat happened
1. Baseline21 campaigns; 42% of spend in Display and Demand Gen; broad match unmanaged
2. Rebuild begins11 restructured Search campaigns launched; legacy campaigns wound down; PMax conversions scaled
3. Legacy offAll legacy Search and Display switched off; PMax reaches 58% of conversions
4. Scale and testEU product PMax launched; bid experiments opened; Demand Gen enters its final stage
5. ConsolidationDemand Gen fully retired; global financial services campaign added; bid tests judged and cut
6. Steady state16 campaigns (12 Search + 4 PMax), zero broad match serving

Performance Max vs themed Search

Performance Max buys conversion volume cheaply but broadly: 2.5–3.5% of its leads were rated High, versus 13–25% from themed Search. PMax fills the funnel; Search supplies the qualified conversations. The budget split between them is set deliberately and reviewed.

Share of Leads Rated High by Sales, by Campaign Type

View the numbers behind this chart
PMax vs themed Search
Campaign typeRoleHigh-rated lead share
Performance MaxFill the funnel; 70% of conversions on 22% of spend2.5–3.5%
Themed SearchSupply qualified, sales-ready conversations13–25%
The scale challengeAs PMax scaled, low-quality leads from universities and NGOs began flooding the CRM.
The fixPMax was set to optimise only to medium- and high-threshold leads, containing the junk without losing volume.

The account before and after

The account moved from 21 blended campaigns across six channel types to 16 campaigns across two, where every live campaign maps to one product theme in one region.

Account comparison
AreaBeforeAfter GrowthSpree
Success metricPlatform cost per leadSales-rated lead quality and attributed deals
Channel mix6 channel types; 42% of spend on Display + Demand GenSearch + Performance Max only
Match types71 broad keywords, 31% of Search spendPhrase and exact only; brand isolated on exact
Campaign structure21 campaigns blending geo, theme and funnel16 campaigns, one theme per region each
Landing pages26 pages incl. docs and expired webinars; 36% of spend on product pages16 pages; 79% of spend on theme-matched product pages
Performance Max5% of conversions70% of conversions on 22% of spend, quality-thresholded
Documentation pages taking clicks71

Same cost per lead, much better leads

Overall cost per lead stayed roughly the same, but the leads were materially better: high-intent share rose from 1.6% to 5.3%, low-quality share fell from 82% to 69%, and deals followed the restructured build.

Before vs After: Seven Numbers That Moved

View the numbers behind this chart
Before vs after
MetricBeforeAfter
High-intent (High-rated) share of paid leads1.6%5.3%
Low-quality share of paid leads82%69%
Restructured campaigns' share of attributed deals26%61%
Restructured campaigns' share of attributed pipeline value42%67%
Spend on product landing pages36%79%
Display + Demand Gen share of spend42%0%
Broad-match share of Search spend31%0%

DataHub's results

1.6% → 5.3%High-intent share
of paid leads
−50%Cost per
high-quality lead
+79%Deals from
restructured campaigns
5% → 70%Performance Max
share of conversions
42% → 0%Display + Demand Gen
spend
31% → 0%Broad-match share
of Search spend

What didn't work, and what we did about it

Not everything worked first time. These are the calls that shaped the result.

  1. Reported cost per lead went up.Once cheap content downloads stopped counting, platform dashboards looked worse for a while even as lead quality climbed. We reported on sales-rated leads so the change wasn't misread.
  2. Scaling Performance Max brought in junk.Low-quality leads from universities and NGOs started flooding the CRM. We fixed it by optimising PMax only to leads that met medium- and high-quality thresholds.
  3. Several bid experiments didn't beat the control.They were judged and cut rather than left running.
  4. Demand Gen couldn't be switched off overnight.It was wound down in stages so the account didn't lose volume while restructured campaigns ramped up.

Why it worked

Judge channels on CRM lead quality first

A cheap conversion that never becomes a conversation is not a win. Every channel is measured on sales-rated leads first.

Structure makes optimisation possible

Nothing was diagnosable in the blended account. Bifurcation is why any single theme can now be funded or cut on its own.

PMax and Search do different jobs

PMax buys volume efficiently; themed Search supplies the qualified leads. The split is set deliberately, and reviewed.

Optimise for pipeline

Reported CPL is an input. Sales-rated quality and deals are the outcome, and that is what we manage to.

How to rebuild B2B SaaS Google Ads for pipeline

If your Google Ads account reports a healthy cost per lead but sales says the leads are weak, this is the sequence we used for DataHub.

  1. Score every channel on CRM lead quality

    Pull sales lead ratings and deal attribution for each campaign type before trusting platform cost per lead.

  2. Retire channels that produce no qualified pipeline

    Wind down Display and Demand Gen in stages when their conversions are content downloads that never become sales conversations.

  3. Enforce match-type discipline

    Remove broad match, run phrase and exact split by intent, and isolate brand on exact match.

  4. Bifurcate campaigns by theme and region

    Give each product theme in each region its own campaign, budget, keywords and landing page so it can be measured and funded independently.

  5. Send paid clicks to product pages

    Move traffic off documentation, blogs and expired webinars onto theme-matched product pages you control and test.

  6. Use Performance Max for volume, Search for quality

    Let PMax fill the funnel, optimise it only to medium- and high-quality leads, and keep themed Search for qualified conversations.

  7. Manage to deals and pipeline

    Treat cost per lead as an input and review budget splits against sales-rated lead share and attributed pipeline.

Frequently asked questions

GrowthSpree rebuilt DataHub's Google Ads account around qualified pipeline. The share of paid-search leads rated High by sales rose from 1.6% to 5.3%, cost per high-quality lead fell 50%, deals from the restructured campaigns grew about 79%, and restructured campaigns rose from 26% to 61% of attributed deals and from 42% to 67% of attributed pipeline value.

Judge every campaign on sales-rated lead quality in the CRM rather than platform cost per lead, retire channels that only produce content downloads, remove broad match, split campaigns by product theme and region, and send clicks to product pages. For DataHub, a B2B data and AI SaaS, this lifted high-intent lead share from 1.6% to 5.3% and grew deals from the rebuilt campaigns by 79%.

GrowthSpree, a B2B SaaS and B2B tech paid acquisition agency specialising in Google Ads, LinkedIn Ads and Meta Ads, rebuilt DataHub's Google Ads program. GrowthSpree optimises toward CRM-rated lead quality and pipeline rather than platform cost per lead.

DataHub is a metadata platform for data discovery, data governance, data lineage and data observability, used by data teams to understand and manage their data assets. Its Google Ads program targets buyers across the US, EU and global markets.

Most of the cheap leads were gated content downloads from Google Display and Demand Gen campaigns. They entered the CRM and stopped there. Demand Gen produced zero leads rated High by sales, and Display was credited with deals that carried no pipeline value, so a low cost per lead was measuring form fills, not buyers.

Only if they are judged on CRM lead quality rather than platform conversions. For DataHub, Display and Demand Gen took 42% of spend and looked like top performers on platform metrics, but produced no High-rated leads or pipeline, so GrowthSpree retired both in stages.

71 broad match keywords consumed 31% of Search spend at more than double the account's cost per lead, with no intent control. One broad keyword alone took a third of Search budget. Moving to phrase and exact match only, split by intent with brand isolated on exact, took broad-match spend to zero.

Campaign bifurcation means splitting blended campaigns so each one covers a single product theme in a single region, with its own budget, keywords and landing page. For DataHub, themes such as data governance, data lineage, data observability, metadata and AI, competitor and financial services were split across US, EU and global campaigns, so any theme could be measured, funded or cut on its own.

Performance Max is good at buying conversion volume cheaply but it is broad. For DataHub, PMax grew from 5% to 70% of conversions on 22% of spend and cut its cost per conversion by 61%, but only 2.5 to 3.5% of its leads were rated High, versus 13 to 25% from themed Search campaigns. GrowthSpree uses PMax to fill the funnel and Search to supply qualified conversations, and optimises PMax only to medium- and high-threshold leads.

When scaling PMax for DataHub started bringing in low-quality leads from universities and NGOs, GrowthSpree contained it by optimising PMax only toward leads that met medium- and high-quality thresholds, instead of every form fill.

Paid clicks moved off developer documentation, expired webinars and blog posts onto product pages matched to each campaign theme. Spend on product pages rose from 36% to 79%, distinct paid landing pages fell from 26 to 16, zero-conversion pages fell from 12 to 5, and wasted landing-page spend dropped from 4.8% to 1.2%.

Is your cost per lead hiding weak pipeline?

We tripled DataHub's high-intent lead share by optimising for the conversions sales could actually work. Let's see what your account is really buying.

Get Your Free Google Ads Audit