Lead Segmentation Fix That Unlocked 1,800% More MQLs

Check how the lead is defined before adding more of them. At Semrush an MQL required a business email domain, so anyone who signed up with gmail, yahoo, aol, or outlook was filed as unqualified and never routed. Most free-trial signups use exactly those addresses, or temporary ones, because they are harder to attribute back to a buyer.

Auditing new signups surfaced the pattern the rule was missing. Semrush served a large base of small business owners, and those owners put the company name before the @, not after it: joesgaragedoors@gmail.com. Searching those addresses matched real business profiles on Facebook, where SMB owners actually are, rather than LinkedIn. The enrichment logic was reading only the domain and throwing the company name away.

Rewriting it to read the local part of the address as a company signal released +1,800% more MQLs into the correct paths. Demand was never the problem. The definition was.

A single-product funnel breaks the moment the catalog grows. Semrush was expanding from a core product to more than 50 internal products plus an app store, and no one had ever attempted a shared cross-funnel view. Each team reported on its own slice, so nobody could say where a customer entered, what they adopted next, or where they leaked out.

The fix was a single map covering Attract, Engage, Convert, Upsell, and Churn, built once and used by every team, across NA, APAC, EMEA, and LATAM.

A first of its kind at the company: one end-to-end view of Attract to Churn, built in Looker Studio and still in use today. It carried prediction modeling on return rates, cancellation reasons by source, lead scoring buckets by channel and country, a campaign-level filter, and the top five retention tools for free versus paid users.

The test for an executive dashboard is whether it changes a decision in the meeting it is shown in. This one set the shared definition of the funnel that the rest of the work was built on.

The budget layer on top of the funnel. Spend was going out shotgun-style across every channel in the hope something stuck. Mapping the journey by channel and by stage made it methodical: which channel, at which stage, owned by which team, at which point in the customer journey.

That turned budget allocation into an argument about evidence rather than volume.

Enrichment ran through Clearbit, since acquired by HubSpot. The rebuild picked the data points that actually decide routing with analytics, demand generation, product, billing, and sales enablement in the room, then had engineering build a data architecture that could carry them at volume.

On top of that sat automated routing by industry, company size, and product behavior, lead scoring, ABM tracks running against both MQLs and SQLs, and a data re-activation motion for dormant records, plus standardized intake templates for webinars, ebooks, and forms so new leads arrived in one shape instead of twelve. Together the lifecycle architecture lifted visitor to trial velocity by 20%, visitor to new MRR by 10%, and cut churn by 10%.

Paid and free lookalike audiences, built by analyzing closed-won customers next to freemium users to isolate what actually separates someone who pays from someone who stays free. Free signups are not noise in a PLG motion. They are the top of the funnel, and the job is to move them through trust and demonstrated need into paying customers.

The model fed more than ad targeting. It drove product page recommendations and a distinct new-user onboarding page path per track, each designed to reach highest value fastest and build a workflow sticky enough to convert. A segmentation dashboard kept paid, landing pages, and lifecycle email aligned with sales and customer success.

Marketing at Semrush owned everything from anonymous visitor through first payment in a self-serve PLG model, so conversion had to come from internal signals rather than more spend. Implementing FOMO surfaced anonymous customer activity as live social proof on-page, letting real buying behavior influence the next visitor.

It was built with legal, product, marketing, and billing together so the anonymized data use held up to review. It lifted visitor to trial conversion by 20% and trial to first payment velocity by 10%, and worked well enough to be rolled out globally across every landing page and high-traffic page on the site.

Most of the wins above started as a test. Mapping missed opportunities across the funnel produced a backlog that turned into 250+ experiments built from scratch, run as A/B tests against the end-to-end self-service engine: pricing presentation, sign-up flow, onboarding paths, product recommendations, landing pages, and lifecycle email.

The program had two rules. Every test named the decision it would settle before it launched, and losing tests were written up as carefully as winners so the next person did not re-run them. That cadence is where FOMO social proof, the onboarding track split, and the product recommendation logic came from. Internally it earned the nickname New Idea Generator, which is really just what happens when experimentation stops being a quarterly campaign and becomes the operating rhythm.

Co-managing a $500K MarTech budget meant knowing what every tool did and who else was already paying for it. A company-wide audit compared tools across departments, exposed duplicate spend, and unified overlapping systems onto one contract.

The enrichment renewal is the clearest example: comparison matrices and a negotiation took the Clearbit contract to 50% of the original price, $115K saved, with the cost per credit driven down against close to 2.5 million unique signups a year.

Onboarding was slow because knowledge lived in meetings. The fix was a playbook per department plus a playbook per micro-process, documenting how the work actually gets done: goals, roadmap, annual priorities, handoffs, and who to go to.

A new hire, or any team that needed to collaborate with another team, could read instead of scheduling three discovery calls. Onboarding dropped from 90 days to 60 and campaign execution time fell 20%, all centralized as a Center of Excellence with templated campaign briefs, training, and a feedback loop.

Two programs built from nothing. Localization was stood up through Marketo so regional teams could run campaigns in-market rather than translating a US calendar after the fact.

The webinar program moved off Streamyard onto BrightTALK, rebuilt with an internal subject matter expert to lift engagement. It took off, and the SME who inherited it was promoted to Senior Global Campaigns Manager and ran the system for three years after handover. Programs that outlive the person who built them are the real test.

+1,800% more MQLs, unlocked by rewriting how a qualified lead was identified rather than by buying more traffic. Leads that had been discarded as personal-email signups were recognized as small business owners and routed to teams that could work them.

+20% visitor to trial conversion, and +10% on trial conversion velocity, meaning how quickly a trial turns into a first payment rather than just whether it eventually does. Both came from the on-page social proof rollout, later extended site-wide. Social selling added +10.87% visitor to signup, +13.04% visitor to trial, +44.44% visitor to first payment, and +17.19% MRR from first payments, alongside +10% new MRR and -10% churn.

-30% onboarding time, 90 days down to 60, and -20% campaign execution time. Written playbooks by department and micro-process replaced the meetings new hires used to need before they could contribute.

$115K saved on the enrichment contract at 50% of the original price, with a lower cost per credit against close to 2.5 million unique signups a year, plus duplicate tooling removed across departments from a $500K budget.

The first company-wide funnel view across a 50+ product suite, built in Looker Studio, adopted by the C-suite, and still running today. Reporting that held up under acquisition-stage diligence.