Case studies written the way engineers document systems.
Every Bluefrog engagement gets described the same way: what the business actually could not see, which systems held the missing pieces, what we connected, what the AI does, and what management can now decide. No slogans, no invented metrics — the architecture and the outcome.
The Bluefrog case study structure.
Most agency case studies describe a campaign. Ours describe a system. The same eight headings appear every time, because the same eight questions determine whether an AI systems integration is worth building.
Business Problem
Stated in operating terms. What decision could not be made, and what it cost to guess.
Data & Systems Involved
Every platform holding a piece of the answer — ads, phone, forms, CRM, field operations.
Bluefrog Solution
The architecture: what gets collected, normalized, joined, evaluated and surfaced.
Integrations
The actual connections — APIs, webhooks, call data, scheduled syncs, and the keys that join them.
AI & Automation
Where AI does real work: transcription, intent, classification, rubric evaluation, summarization, alerting, follow-up.
Analysis
What the connected data can finally answer — attribution, correlation, trend, efficiency.
Business Result
What changed for management — described qualitatively unless a figure can be substantiated.
Technology
The stack underneath: platform modules, custom services, databases, schedulers, dashboards.
The two cases below are anonymized and representative. They describe engineering patterns we build repeatedly — not a single named client.
From lead generation to revenue intelligence.
A multi-technician home services operation running Google Ads, Local Services Ads, organic search and a heavy inbound phone volume — with ServiceTitan as the system of record for everything that happened after the call.
Four systems, four partial truths.
Google Ads reported clicks. Local Services Ads reported charged leads in a separate interface. The phone system held recordings nobody had time to listen to. ServiceTitan held bookings, jobs, estimates and revenue. Each system was accurate about its own slice and silent about everything before and after it.
Management could answer "how many leads" and "how much revenue," but not the question that drives budget: which sources produce revenue, and what is a lead from each one actually worth?
Everything holding a piece of the journey.
| System | What it knows | What it cannot see |
|---|---|---|
| Google Ads | Impressions, clicks, cost, conversion events | Whether the job was booked, completed or profitable |
| Local Services Ads | Charged leads, calls, message leads | Which leads became estimates or revenue |
| Phone system | Call time, duration, recording | Intent, quality, whether an opportunity was missed |
| Website | Sessions, sources, form submissions | What happened to the customer afterward |
| ServiceTitan | Bookings, jobs, estimates, sales, revenue | Which marketing activity originated the customer |
One record per customer journey.
Bluefrog built the connective layer. Advertising activity, call records, website sessions and lead submissions are collected continuously and joined to the ServiceTitan record of what the business actually did — booking, dispatch, job, estimate, sale, revenue. One journey, one row, queryable by source, by service, by location and by date.
API-level, not spreadsheet-level.
Google Ads and Local Services Ads APIs, Google Business Profile, call and recording retrieval, website and form capture, and the ServiceTitan API for bookings, jobs, estimates and invoices. Scheduled syncs plus event-driven updates, with reconciliation so late-posting revenue lands on the right day.
The calls stop being a black box.
Every call is transcribed, summarized and classified for intent and bookability by the voice intelligence module, then scored against the company's own CSR rubric. Automation pushes context back where it is useful and raises alerts when a qualified opportunity was not booked.
Questions that were previously unanswerable.
Lead volume and revenue on the same timeline. Revenue per lead by source. Booking rates by source, service and team. Daily and weekly trend. Marketing efficiency — what each channel returns rather than what it costs. This is the working definition of revenue intelligence.
Marketing decisions stopped being opinions.
Management moved from four disconnected reports to a single view of lead volume, revenue, revenue per lead, booking rates, source performance and trend. Budget conversations shifted from cost per lead to revenue per lead, missed opportunities became visible the same day, and hand-assembled reporting time went back into operating the business.
Platform modules plus custom services.
Bluefrog Intelligence Platform modules for voice, marketing and revenue intelligence, extended with custom ingestion services, a normalized data store, scheduled reconciliation jobs and custom management dashboards. Where the platform covered the pattern, we configured it. Where it did not, we built it.
Growing revenue while reducing cost.
The instinct when revenue needs to grow is to spend more. The pattern we keep building does the opposite: connect the systems, automate the manual work, and let better information reallocate the money that is already being spent.
Results reflect selected client engagements. Individual results vary based on market, company, implementation and other factors.
Why the two move together.
Revenue and cost are usually treated as a trade. They stop being one when the waste is informational rather than operational — money spent on channels that generate leads but not revenue, hours spent assembling reports a system could produce, opportunities lost because nobody could listen to every call, duplicate entry between disconnected tools.
Connect the systems and those costs come out while revenue improves, because the same connected data that removes the waste also shows where demand actually is. That is the argument for Operational AI: instrumentation, not a creative tool bolted onto the side of the business.
Shape of the pattern only — indexed to 100 at start. Not client data.
Growth targets rising while budget scrutiny increased. Spend allocation was based on platform-reported cost per lead, and a meaningful share of management time was consumed producing reports rather than acting on them.
Advertising platforms, call and communication records, website and lead capture, CRM and field operations data, plus the manual spreadsheets that existed only because those systems did not talk to each other.
API integration across marketing, communications and operations; AI transcription, classification and rubric evaluation; automated reporting, alerting and follow-up replacing recurring manual work. See automation intelligence.
Spend-versus-revenue by source, efficiency trend and pipeline exposure surfaced through an AI business analyst layer over a connected data store, with dashboards for each level of management.
Grow revenue while improving efficiency.
Every efficiency engagement gets decomposed into four areas. We measure each one separately, because they fail separately.
Marketing Efficiency
Budget follows revenue instead of cost per lead. Channels that generate volume without booked work become visible, and allocation shifts toward sources that produce jobs.
Employee Efficiency
AI handles the reading, listening, summarizing and logging. People handle customers and decisions. Evaluation against your own standards makes coaching specific instead of general.
Management Efficiency
One connected view replaces the weekly assembly of numbers from five interfaces. Leadership spends its time interpreting the business rather than reconstructing it.
Technology Efficiency
Connected systems eliminate duplicate entry, redundant tools and the custom exports that hold everything together. Integration usually costs less than the workarounds it retires.
Illustrative scoring format — an efficiency review scores these four areas for your operation before any build begins.
The first document we write is yours.
Before anything is built, we write the first three sections of your case study: the business problem in your words, the systems holding the pieces, and the architecture that would connect them. If those three sections do not describe something worth building, the engagement should not happen. Most work for home services operators follows the first pattern; most multi-location and mid-market work follows the second.
AI is easy to access. Making it useful is hard.
Bluefrog makes AI useful by integrating it with the way your business actually works — your software, your calls, your customers, your marketing and your revenue.
Technology development since 1997 · AI integration platforms since 2001