What the work actually returned.

Situation

~€25k/mo on LinkedIn lead-gen forms. Strong MQL volume, but <3% converted to SQL and close rate was 2%. Sales thought marketing was sending noise; marketing thought sales wasn't calling back. Both were partly right.

What we changed

• Killed all lead-gen forms. Every campaign now clicks to a real landing page.
• Rebuilt offers around demo + gated teardown (not ebooks)
• Wired HubSpot enhanced conversions with stage-based values
• Added competitive + warm-audience retargeting

What happened

• Leads/mo fell from ~180 to ~55. Sales stopped complaining.
• SQL rate went from 3% to 14% over two quarters
• Pipeline sourced from LinkedIn: 3.2× vs baseline quarter
• CAC down roughly 40% despite higher CPL

Situation

Running Google Search only, single flat campaign mixing brand, category and competitor queries. Three competitors bidding on the brand term. PMax had been rejected internally as "unpredictable". Sales thought Google wasn't working; they couldn't see it was doing most of their verification.

What we changed

• Split into six campaigns: brand defence, competitor, category, bottom-funnel, PMax-audiences, YouTube remarketing
• Wired offline conversions from HubSpot closed-won
• Killed all broad match outside brand
• Rebuilt ad copy around differentiation, not features

What happened

• Brand CPC fell 48% once we outbid competitors on defence
• PMax-audiences drove 22% of net-new pipeline in Q2 post-launch
• CAC from Google dropped 34% over 9 months
• Share of branded SERP clicks climbed to 81% (from ~50%)

Situation

Founder-led sales, 40 inbound demos/month, all word-of-mouth. Wanted to start paid without blowing the CAC maths and without becoming dependent on it. Google SERP in their vertical was brutal, three established players with 10-year-old brand campaigns. LinkedIn-addressable ICP was tight (~18k contacts in the UK).

What we changed

• Skipped Google Search entirely, CPCs were €40+ and we'd bleed
• Started LinkedIn + Microsoft Ads on demand-capture only
• Built an ICP list of 18k contacts; ran Sponsored + Message Ads
• Every campaign had a hard CAC ceiling (€8k) before auto-pause

What happened

• Month 1: nothing. Month 2: first 3 SQLs. Month 3: 11 SQLs.
• €180k sourced pipeline in Q2 vs zero baseline
• Blended CAC held at ~€6,800, below the €8k ceiling
• Found 2 new ICP segments we didn't know were active buyers

Situation

Two agencies before us, €80k/mo on paid across LinkedIn + Google + Meta + Microsoft. No reliable pipeline attribution, HubSpot saw everything as "direct" or "LinkedIn" regardless of source. Previous agency had been asked to scale spend; they did, and CAC doubled quietly.

What we changed

• Stopped every scale recommendation. Spent month 1 on measurement only.
• Rebuilt GTM with server-side + consent mode + offline conversions
• HubSpot-to-ad-platform sync rewired from scratch
• Killed 4 of 7 campaigns based on actual-pipeline data

What happened

• Real CAC was 2.3× what the previous dashboards showed
• Cut spend from €80k/mo to €56k/mo over one quarter
• Pipeline rose 45% year-over-year on 30% less spend
• Attribution confidence, measured by sales + finance alignment, finally unblocked budget increases

Situation

AI software sold to large enterprise contact centres. A buying group of CX, operations, IT and executive roles, sales cycles of six months and more, and pipeline tracked in Dreamdata and Salesforce.

What we changed

• Built a full-funnel account-based structure with cold, warm and hot layers. Target accounts that had seen few or no ads were used to push budget toward accounts not yet reached.
• Excluded existing customers and partners from every campaign.
• Added the Dreamdata pipeline stages, from MQL to closed-won, as conversions on every active campaign. The client's team ran the Dreamdata and Salesforce connection; we ran the LinkedIn and web conversion setup.
• Split the broad cold campaign by persona: CEO, Head of CX, technology leaders and heads of AI transformation.

What happened

• USD 675,167 in attributed pipeline over six months (November 2025 to April 2026), on USD 125,754 of LinkedIn spend.
• Most of it came from the warm, account-based campaigns aimed at accounts already engaged.
• 73 influenced opportunities and 7.6 attributed opportunities in Dreamdata's data-driven model.

Situation

An API product sold to developers and product teams. The buyers were engineers, engineering managers and product managers at a defined list of target companies, and the budget had to reach them without paying for everyone else.

What we changed

• Built the target list from the client's HubSpot, with existing customers excluded.
• Ran three layers instead of one broad campaign: one-to-many ads to the top-100 target list, one-to-one ads for 13 named accounts, and remarketing to people who had already engaged.
• Checked delivery by job title as well as by company.

What happened

• All 13 one-to-one accounts reached, at a 1.42% click-through rate.
• 67.7% of all impressions landed on companies on the target list.
• Job titles reached included Product Manager, Senior Product Manager, Engineering Manager, Director of Engineering, Head of Engineering and Lead Engineer.
• Eight accounts with open deals worth USD 820k in total had seen the ads before each deal was created. That is influenced pipeline; none had closed when the project ended.

Situation

A UK marketplace connecting consumers with financial advisers. We ran LinkedIn to the business side: advisory firms and wealth managers. LinkedIn's attribution report and the client's HubSpot told different stories: of the first 12 deals LinkedIn claimed, none carried LinkedIn as a traffic source in HubSpot.

What we changed

• Matched every deal the platform claimed against the CRM, deal ID by deal ID, before reporting any of it.
• Separated monthly deal value from total contract value, which the CRM holds in different fields.
• Traced the gap to its cause: a hosted form was stripping the LinkedIn source, so LinkedIn traffic showed up in HubSpot as direct.

What happened

• 27 LinkedIn-attributed deals verified: 16 closed-won and 11 open (1 January to 10 June 2026).
• GBP 2.0 million of verified pipeline in total contract value.
• The closed-won deals alone came to GBP 46,909 in monthly deal value against GBP 36,019 of ad spend, before any of the open pipeline.

Situation

Cloud point-of-sale software for restaurants, retail and service businesses, part of a European software group. When we audited the account in March 2026, it was losing ground: conversions per day down 27% and cost per conversion up 25% over six weeks. Leads imported from Salesforce were under-counted and valued at EUR 30.

What we changed

• Rebuilt the measurement: lead value raised from EUR 30 to EUR 380, and generic search switched to bid on Salesforce leads only.
• Ran a 130-day experiment of broad match against phrase and exact. On head terms the cost per lead came out even, and broad brought more volume. On five niche ad groups, phrase and exact cost less per lead.
• Added competitor campaigns in Germany and Austria.
• After a statistical test, paused 102 keywords with no conversions of any kind in 180 days. Kept 16 that had no leads but had produced other conversions, including closed deals.

What happened

• Cost per lead, non-brand search

  Leads counted in Salesforce, as read on 1 October 2026. Salesforce adds late leads for up to 90 days after a month closes.
• July: 183.3 leads against 144 planned, at EUR 356 per lead.
• August: 198.8 leads against 177 planned, at EUR 361 per lead.
• Both under the EUR 386 target, on EUR 65k to 72k a month of non-brand search.
• Over 90 days, the Austrian competitor campaign produced 70.9 leads at EUR 309 each.

Situation

Sales intelligence for companies that sell to the UK public sector. LinkedIn had mostly run gated reports and playbooks behind lead forms, which brought in leads cheaply. Deals sit in HubSpot, which is connected to LinkedIn's revenue attribution report.

What we changed

• Built the cold layer on the client's HubSpot target-account list combined with the buyer job titles.
• Ran it on website visits, thought leader ads, spotlight ads and text ads, with a lead form on one cold campaign only.
• Added a warm layer retargeting website visitors and people who engaged with the ads, including a conversation ad.
• Handed the build over in a training session. The client's team has run it since April.

What happened

• USD 771k of open pipeline credited to the new campaigns in LinkedIn's revenue attribution report, 1 April to 24 September 2026, on USD 41k of spend.
• USD 82.6k closed-won credited to the same campaigns in that period.
• Seven of the eight campaigns we built were still running under the client's team six months after launch.