Case studyCustomer-conversation SaaS
Name under NDA
EUR1.8 million in additional ARR for a customer-conversation SaaS company
- EUR1.8M
- 530
- +61%
Who the client is
The business is an EU-based software company serving approximately 1,500 paying businesses at kickoff. Its shared inbox combines email, WhatsApp and social messages, routes conversations, supports automation and passes requests to staff. Customers include distributors, wholesalers and travel and hospitality businesses.
The commercial model has three subscription tiers varying by team size and usage. The revenue calculation uses period-average monthly subscriptions of EUR450 per trial-route company and EUR850 per demo-route company, excluding messaging fees. Our engagement covered English and German acquisition.
What they came to us with
The client wanted more suitable paying businesses from search and fewer poorly matched demonstrations. Buyers needed to understand whether the software could handle their conversations, integrations and costs before choosing a trial or speaking to sales.
Their visibility goals included AI recommendations that explained supported workflows and limitations. The marketing team also needed a sustainable way to keep multilingual product guidance current as features and platform policies changed.
What they got
The completed engagement ran from September 16, 2025 to September 15, 2026. At the end, customers acquired through Google organic search represented an estimated EUR1.8 million in additional annual recurring revenue (ARR), above the amount expected without the engagement. Revenue earned during the year was measured separately.
- EUR691Kexpected without the engagement
- EUR144Kcovered our fees
- EUR326Kgained on top
AI answers improved across the same 120 non-branded buying questions, covering both buyer segments and languages. ChatGPT, Perplexity, Gemini and Claude were sampled on three days per measurement, producing 1,440 answers each time. Answers linking to the client’s website rose from 108 to 288. The share of assessed product claims matching approved facts rose from 72% to 92%.
September 2025September 2026
Answers linking to the client websiteOf 1,440 sampled answers
Share within selected brandsClient appearances among all appearances of the client and four competitors: 180 of 1,500, then 360 of 1,800
Favorable sentimentAmong answers mentioning the client
Product precisionCorrect claims among assessed claims
Agency fees totaled EUR144,000 across twelve full monthly payments of EUR12,000.
The situation before the engagement
Business context
Buyers needed specifics when comparing the client with messaging specialists and larger support platforms. Trials served simpler requirements; demonstrations helped assess more complex deployments. Conversation ownership, integration requirements, human handoff and usage costs all affected the decision.
Website and content
The website contained approximately 4,700 content pages across six languages, including 1,650 English/German pages. Of those, 1,500 editorial and commercial pages were eligible for this work. Each language version counted separately; redirects were excluded. Product, comparison, integration and advice pages existed, but overlapping guides competed and revisions lagged behind product and policy changes.
Traffic sources
Acquisition traffic totaled approximately 900,000 sessions across all languages in the preceding year. About 800,000 reached English/German content, including 360,000 Google organic sessions. Direct visits, partners, paid acquisition, email and social supplied other traffic. Product logins and help-center visits were excluded; identifiable AI referrals were counted separately.
- 360,000Google organic, English and German
- 440,000Other acquisition channels, English and German
- 100,000Other languages
Search presence
Existing Google rankings provided a foundation, but comparisons and workflow questions were underrepresented. AI answers emphasized generic chat automation and often omitted supported channels, integration dependencies and handoff requirements. Buyers received an incomplete picture of the product.
42 in the top ten at kickoff
78 a year later
- Google top three
- Positions four to ten
- Below the first page
The same 240 buyer queries, one square each.
Previous approach
The marketing team published messaging advice and feature announcements, asking product specialists for information as needed. Teams handled old claims, translation updates and buyer objections individually. Website engineers implemented changes.
The strategy we planned and why
Commercial and search priorities
Specific buying decisions guided our SEO & AI Search Strategy. A wholesaler needed to collect quantities and delivery requirements before assigning a quote inquiry. A hotel needed the next shift to see earlier guest messages and pass complex requests to a person.
Clear product boundaries mattered as much as features. We planned explanations of when the inbox fit and when a CRM, dedicated ticketing system or custom integration was needed. Smaller teams could prepare for a trial; complex prospects could arrive at a demonstration with specific requirements.
Technical SEO priorities
Technical priorities included rendered commercial content, titles, canonical URLs, language mappings, internal links and stale redirects so search systems could reach the correct English or German explanation. These fixes supported the same commercial pages prioritized by the strategy.
Content strategy and production order
We prioritized corrections to commercial pages, followed by comparisons, channel guidance, integration requirements, setup and costs. Content Research & Strategy established this order. AI Search Optimization (GEO & AEO) used approved product facts to check website claims and generated answers, including limitations and human escalation.
The production system connected documentation, demonstrations, support material and expert interviews to drafting, review and bilingual publishing through our AI Content Production Systems service. Weekly meetings taught the client team to apply the strategy, with Slack discussions between calls.
Why the strategy fit
Existing content offered earlier opportunities while new coverage developed. Competitor research examined quote qualification, guest continuity, integrations and costs. Broad messaging articles left these objections unanswered. English and German had commercial relevance and assigned reviewers.
What we did
Agency and client responsibilities
- We led strategy, research, technical specifications, drafting, production-system tuning and verification.
- Client engineers implemented website changes; product specialists approved claims; English and German reviewers checked language and suitability.
- Client sales and success teams ran demonstrations, trials and onboarding. Weekly reviews connected their questions and AI-answer errors to publishing priorities.
- 1Research, strategy and technical fixes
- Buyer research
- Technical audit and fixes
- Approved product facts
- Strategy and plan
- 2AI content production system
- Drafts from approved sources
- Fact checks and product approval
- English and German review
- 3Ongoing production, measurement and improvements
- New English and German articles
- Updates to existing pages
- Corrections when features change
Weekly reviews with the client team, adjusting priorities from sales questions and AI errors
Phase 1: Research, strategy and technical fixes
The first month, September 16 to October 15, 2025, combined buyer research with an audit. We prioritized issues by commercial impact, approved product facts, agreed on responsibilities and completed the strategy and 90-day plan.
Technical work included:
- Improving indexable service and integration content and rewriting unclear titles and descriptions so buyers and search systems could identify the page’s purpose.
- Correcting canonical, redirect and language relationships and reviewing server logs and broken crawl paths so discovery reached the intended version.
- Connecting use-case, integration and comparison pages to help visitors follow a buying question through to requirements.
- Checking organization, software and breadcrumb structured data against page content.
We verified implementation by recrawling affected templates, comparing rendered pages and checking Search Console coverage and Google results. These checks confirmed initial fixes across several issue groups.
Phase 2: Building the AI content production system
The second month, October 16 to November 15, focused on building and testing the marketing-content production system. We configured it to:
- Draft from approved documentation, demonstrations, support material and expert interviews.
- Check facts and queue product approval, followed by English or German language review before publishing.
- Maintain links between source facts and published language versions so changed features or policies could trigger corrections.
Product specialists verified whether claims described native functionality, integration dependencies or human action. Testing preceded new article production.
Maintenance remained active. Existing-page corrections continued during system setup, alongside technical monitoring and release checks.
Phase 3: Ongoing production, measurement and improvements
Production ran from November 16, 2025 to September 15, 2026. Weekly reviews adjusted priorities using sales questions and AI errors. Output varied with complexity and product review.
The resulting inventory was approximately 4,800 retained content pages. Updated-page counts describe distinct retained pages; consolidated sources became redirects. The 200 originals comprised 120 English and 80 German articles. Shared technical templates could affect other languages; content production covered English and German.
200 original articles
80 localized versions
700 retained pages updated
180 overlapping pages merged
One square is 10 pages.
Results
Search results
- English/German Google organic sessions, 2025-2026
- Monthly average the year before
- Buyer queries in Google's top ten
- 4278
- of the same 240 queries
- Buyer queries in Google's top three
- 1227
- of the same 240 queries
Business results
Trial entries per session rose from approximately 0.5% in the first engagement month to 0.9% in the final month. Trial and demo entry ratios divide Google-attributed company entries by English/German Google acquisition sessions, including repeat visits. These channel-level ratios compare company counts with visits. Acquisition sessions exclude product logins and help-center use.
Trial activation rose from approximately 50% to 60%, comparing the first cohort with July 16-August 15 entrants. Paid starts followed trial entry by approximately one month. Activation required a connected channel, an invited teammate and ten genuine conversations within fourteen days. Activated-to-paid conversion remained approximately 20% for these mature cohorts.
Companies entering a trial from English/German Google search
Year before
Engagement year
Companies from both routes starting paid plans before each cutoff
Year before
Engagement year
Demo requests rose from approximately 0.3% to 0.4% of acquisition sessions between the first and final months. Later-stage rates compare the first cohort with June 16-July 15 entrants, whose two-month paid-start window ended before cutoff:
- Held and qualified demos: approximately 50% to 60% of requests.
- Accepted opportunities: approximately 50% to 60% of held, qualified demos.
- Opportunity wins: approximately 20% to 25% of accepted opportunities.
Company-wide growth
Company ARR grew from approximately EUR9.3 million to EUR14.7 million. That EUR5.4 million increase included organic search, other acquisition channels and existing-customer upgrades, after cancellations and downgrades. The estimated additional organic ARR was equivalent to approximately 33% of that increase.
Revenue and ARR from organic search
- Organic-acquired ARR
- Expected ARR without the engagement
- Estimated additional ARR
Additional ARR is calculated from unrounded values. Each displayed amount is rounded separately.
The central estimate allows 10% business-as-usual traffic growth and retains 20% of the improvement at each conversion stage for the client's own work.
The chart follows customers acquired through English/German Google organic search after kickoff. Their first payments arrive after the first monthly cutoff.
Growth built gradually. Additional paying customers began arriving in the fourth month. Earlier customers continued paying as new customers joined, so monthly subscription revenue accumulated through the rest of the engagement.
Conclusion
The client finished with an estimated EUR1.8 million in additional ARR from organic acquisition and EUR470,000 in additional revenue earned during the year. The team also had approved product facts, a working bilingual publishing process and a regular way to use sales questions and AI-answer errors to keep their published guidance current.
How we measured
Search measurement
Traffic comparisons use September 16, 2024 to September 15, 2025 against the engagement year. Rankings compare kickoff with the final report on September 16, 2026. Acquisition analytics, Search Console and Ahrefs supplied traffic and ranking records.
The ranking panel comprised 120 UK-English and 120 Germany-German desktop queries. We counted the best owned organic position, excluding advertisements and AI answers. Sessions count website visits.
AI answer measurement
The AI comparison used answer captures from September 16-18, 2025 and September 13-15, 2026. The 120 fixed questions covered commerce and travel/hospitality in English and German, with 30 per combination. Four engines answered each question on three days. Profound supplied captures; our processing calculated the following measures:
- Each brand counted once per answer. Share of voice compared the client with Respond.io, SleekFlow, Wati and Front.
- Models classified client-mentioned answers as favorable, neutral or negative. Neutral counts were 63 and 72; negative counts were 18 in each measurement.
- Our models checked claims against dated approved facts. Precision measured correctness among assessed claims; completeness of buying guidance remained unmeasured.
Customer attribution and conversion
Company attribution connected acquisition analytics with CRM, workspace and billing identifiers. Inclusion required the first recorded non-direct touch within 90 days before entry to be Google organic on eligible English/German commercial or editorial content. We removed duplicates, existing customers, unresolved identities and pre-existing partner or outbound opportunities.
Route assignment closed at each reporting cutoff. A company requesting both a trial and a demo belonged exclusively to the demo route throughout that report's funnel and revenue totals. Assisted trials without a demo request remained in the trial route.
Annual totals count trial entries and subscriptions begun before cutoff. Conversion rates above use mature cohorts. Outcomes cover acquisition and initial use; poor-fit held demonstrations and sustained workflow use remain unmeasured.
Revenue estimates
Estimated additional results compare organic acquisition with a baseline that allows 10% business-as-usual traffic growth. Each conversion stage retains 20% of its percentage-point improvement for the client's own work. For example, activation rising from 50% to 60% becomes 52% in the comparison. These allowances reflect our judgment of what the client would have achieved alone.
Sensitivity to those allowances gives additional ARR of EUR1.48-EUR2.07 million. The lower estimate allows 20% traffic growth and retains 40% of conversion improvements; the upper allows neither. Both keep route prices, conversion delays and retention fixed. The range covers these two alternative baselines.
Two revenue measures. ARR annualizes the recurring subscription value still active at the end. Earned revenue counts service delivered through September 15, after refunds and cancellations. Annual prepayments were spread across service periods. Both measures exclude taxes, external message charges, the opening customer base and pre-engagement pipeline.
Attribution method. Google acquisition includes branded and non-branded demand. Earlier advertising, partner introductions and offline recommendations can influence those customers. The additional estimate depends on the baseline allowances above. Identifiable AI referrals were counted separately; sampled AI-answer improvements have no assigned share of the Google revenue result.
Rounded figures. Additional ARR is calculated from unrounded values. Each displayed amount is rounded separately.
Become the brand
AI recommends
Book a free 30-minute call with Eugene. Before the call, we check what Google and AI say about you. On the call, you get the first steps to make AI recommend you.