Sergio AlbaBogotá, Colombia

Data EngineeringData AnalyticsApplied AI

Data Analytics, automation and applied AI for marketing teams.

Explore LabsMessage me on LinkedIn

Scroll

01/ 04 · The starting point

Chaos

Every system exports its own way: Google Ads reports spend in micros, Meta uses other field names and the CRM writes dates in its own format. Finance and operations have the same problem with other files. Before analyzing, you have to find what is broken.

Copies · gaps · units · dates · taxonomy

02/ 04 · Data integration and quality

Structure

I integrate marketing, CRM and web analytics sources in BigQuery. I define data structures, document metrics and implement quality checks and reconciliation across sources.

  1. 02.1Repeatable, traceable data loads.
  2. 02.2A documented data model for analysis and reporting.
  3. 02.3Validation and reconciliation rules across sources.

SQL · Python · BigQuery · Google Cloud

See the area

03/ 04 · Process automation and applied AI

Intelligence

I automate recurring processes with Python and SQL: reports, file consolidation, data loads and validations, from extraction to an output ready for review. AI joins the workflow as one more step: it explains variations and proposes a reading that the team reviews.

If the sources change, the rules change; the shape of the workflow stays.

  1. 03.1A documented extraction, validation and output workflow.
  2. 03.2Quality checks and exception handling.
  3. 03.3The team reviews and approves before presenting.

Python · SQL · BigQuery · Power BI · Data Studio · AI in the workflow: OpenAI

See the area

04/ 04 · Campaign and conversion analysis

Decision

I analyze campaigns by channel, audience and funnel stage. I assess spend, costs and the progression from leads to opportunities or sales, and turn it into prioritized recommendations.

  1. 04.1Diagnosis by campaign, channel and segment.
  2. 04.2Conversion and costs by funnel stage.
  3. 04.3Prioritized recommendations and follow-up.

SQL · GA4 · Google Ads · Meta Ads · Data Studio

See the area

AI · analysisPaid Media case · June vs May

  1. Leads600720+20%
  2. Qualified210180−14.3%
  3. Paid Social · qualified9040−50
  4. Search · qualified100120+20
  5. Cost per qualifiedSocial US$250 · Search US$1002.5×

More leads, fewer qualified

The loss is concentrated in Paid Social: it adds 100 leads and loses 50 qualified. Before moving budget, it is worth reviewing the CRM discards in that channel.

Human review · the team validates the reading before presenting it

Fictional data · Reporting Hub, Paid Media case

Applied AIBuild, check, iterate

AI in the workflow,
with human judgment

  1. 01Context

    I structure the context and data needed for an explanation.

  2. 02Model

    AI helps explore data, detect patterns, summarize information and retrieve historical context.

  3. 03Evidence

    AI builds the view; the figures come from the analytical engine and each one can be traced back to its calculation.

    Open Generative Dashboard

  4. 04Human review

    I check the text against the figures, separating facts, hypotheses and recommendations. Anything that requires judgment is not delegated to the workflow.

OpenAI · Claude · Hugging Face · OpenAI Codex · Antigravity

Labs · 4 processes

Four processes you can open and review

Each Lab starts from a concrete problem and shows every step, from the source data to the final reading. The data is fictional; every figure can be traced back to its calculation.

  1. Lab 01Reputation modelAnalytical model
  2. Lab 02Reporting HubAutomation
  3. Lab 03Generative DashboardGenerative AI
  4. Lab 04Marketing data warehouseData engineering

Lab 01/ 04Analytical model

Reputation model

InListening, search, content and survey signalsOutan index you can take apart

Ten signals from social listening, search, content and surveys group into three components and form an index you can take apart down to the signal that moves it.

Open the LabBeverages category · Brand A · fictional data

  1. 01Signals

    Ten sub-metrics from social listening, search, content and surveys, each out of 5.

  2. 02Components

    They group into three components: positioning, perception and consideration.

  3. 03Index

    The three form an index of 3.01 out of 5. At baseline it was 3.30.

  4. 04Reading · baselinefollow-up

    More presence, worse perception

    Mentions almost double, from 4,200 to 7,800, but negative earned-media mentions go from 840 to 3,510. Positioning rises from 3.46 to 3.86; perception falls from 3.75 to 2.78 and drags the index from 3.30 to 3.01.

Lab 02/ 04Automation

Reporting Hub

InCampaign, social, listening and web dataOuta report whose totals reconcile

Reports on campaigns, social media, social listening and web analytics, built from the source data. Each one reconciles its totals before proposing a finding.

Open the LabJune 2026 · fictional data

  1. 01Paid MediaReconciles
  2. 02Social MediaReconciles
  3. 03Listening + SearchReconciles
  4. 04Web AnalyticsReconciles

Paid Media · June vs MayTemplate · Campaign / Audience Analysis

12,000 + 10,000 + 5,000 + 3,000 = US$30,000 · all 4 channels reconcile to the total · D+7 CRM rule

More leads, fewer qualified

720 leads (+20%) and 180 qualified (−14.3%). CPL falls, but cost per qualified lead rises to US$166.67.

Lab 03/ 04Generative AI · PromptBI

Generative Dashboard

InA written questionOuta verifiable dashboard

You write what you need to see and the dashboard recomposes itself. AI builds the view; the figures come from the analytical engine and each one can be traced back to its calculation.

Open the Labmarketing_demo v1.0.0 · synthetic data

›01 / 07

 

Trace · SQL

Lab 04/ 04Data engineering

Marketing data warehouse

InRecords from three platformsOuta fact table with no duplicates

Six records, from the file received to the table the marketing team will use. Two arrived duplicated.

Open the Lab30 Jun 2026 · USD · sample data

  1. 01Ingestion · staging

    Google Ads, Meta Ads and LinkedIn Ads deliver 6 records in 3 formats. They land in the raw layer as they come.

  2. 02Transformation

    Same fields and types; micros become USD. Units are normalized; currencies are not converted.

  3. 03Quality · deduplication

    Each row’s key is its grain: platform, account, campaign and day. Two records repeat the key; the load merges and creates no new rows.

  4. 04Dimensional model

    A fact table with 4 valid rows and date, platform, account and campaign dimensions, ready for reporting.

File totalUSD 5,200

Career · Aug 2019 — today

From monitoring conversations to automating marketing data

Sergio Alba sitting at his laptop, wearing a dark shirt
Sergio Alba

Four companies and six roles. The path runs from social monitoring and conversation classification to integrating, validating and automating data at scale.

Aug 2019 — Aug 2020 · BRM S.A. · Keeper RRSS

  1. BRM S.A.

    3 yrs 6 mos at BRM
    Aug 2019 — Aug 2020

    Keeper RRSS

    Social monitoring, reputation alerts and conversation classification.

    View detailsHide details: Keeper RRSS, BRM S.A.

    1 yr 1 mo · Part-time · Bogotá, Colombia

    • Monitored social media and grouped conversations by sentiment, impact and segment.
    • Produced alert reports to track relevant conversations.
    • Handled customer requests and identified opportunities to improve and automate service processes.

    Tools and topicsSocial monitoringSentiment analysisReputation analysisDigital customer service

    Sep 2020 — Aug 2022

    Trend Analyst

    Digital intelligence for strategy, reputation and crisis response.

    View detailsHide details: Trend Analyst, BRM S.A.

    2 yrs · Full-time · Colombia

    • Monitored conversations, posts, news, blogs, forums and influencers to identify trends across the digital ecosystem.
    • Developed real-time dashboards and reports to track digital activity.
    • Analyzed impact, reach, segments and KPIs to inform strategy and crisis management.

    Tools and topicsSocial listeningTrend analysisDashboardsCrisis monitoring

    Aug 2022 — Jan 2023

    Data Analyst

    ETL pipelines, statistical analysis and Power BI for social data at scale.

    View detailsHide details: Data Analyst, BRM S.A.

    6 mos · Full-time · Colombia

    • Executed ETL processes on large social media datasets, ensuring data integrity and consistency.
    • Applied statistical analysis to identify patterns and turn social data into useful findings.
    • Developed Power BI dashboards to present results to teams, stakeholders and decision-makers.

    Tools and topicsETLStatistical analysisPower BI

  2. McCann Worldgroup

    1 yr 4 mos
    Jan 2023 — Apr 2024

    Senior Data Analyst

    Data, social intelligence, automation and applied AI for brand decisions.

    View detailsHide details: Senior Data Analyst, McCann Worldgroup

    1 yr 4 mos · Full-time · Hybrid · Colombia

    • Led digital data analysis and social listening, with real-time monitoring of trends, brand perception, competitors, risks and opportunities.
    • Turned unstructured data and consumer behavior findings into insights for marketing, communications and brand positioning.
    • Combined social listening, performance data and market signals in reports and Power BI dashboards, and applied AI to marketing and analytics workflows.
    • Automated and improved ETL processes across Emplifi, Meltwater and Semrush, including data collection workflows built with Python and Flask.

    Tools and topicsData analysisSocial listeningETLPythonFlaskPower BIApplied AI

  3. Sancho BBDO

    1 yr 5 mos
    Apr 2024 — Aug 2025

    Performance Lead

    Performance leadership, SQL analysis and clear optimization priorities.

    View detailsHide details: Performance Lead, Sancho BBDO

    1 yr 5 mos · Full-time · Hybrid · Colombia

    • Led performance analysis and campaign optimization strategy using data across channels, platforms and accounts.
    • Used SQL to investigate performance changes and turn findings into recommendations for internal teams and clients.
    • Developed reporting frameworks to compare channel, audience and creative performance and establish improvement priorities.
    • Collaborated with media teams, analysts and clients to translate business objectives into measurable outcomes.

    Tools and topicsSQLData analysisPerformance

  4. Publicis Groupe

    Current
    Aug 2025 — today

    Senior Data Analyst

    Marketing data integration, validation and automation for decision-focused reporting.

    View detailsHide details: Senior Data Analyst, Publicis Groupe

    Full-time · Remote · Colombia

    • I integrate marketing data from multiple platforms and use SQL and BigQuery to clean, standardize, validate and structure information.
    • I automate analytics, reporting, validation and reconciliation workflows with Python and SQL to support scalable operations.
    • I develop quality checks to identify metric discrepancies, missing entities, taxonomy errors and inconsistencies between source data and analytical outputs.
    • I analyze digital marketing performance and add context to reports to explain changes, risks and next steps.
    • I collaborate across teams to investigate performance, resolve data issues and turn findings into clear recommendations.

    Tools and topicsSQLBigQueryPythonData qualityAutomation

Tools and technologies

39 tools across 7 areas

Technologies I have used in data integration, analytics, automation, AI and application development. Select a category to see how I apply them.

01 / 07 · Reconciliation & data quality

Data & cloud

Turn scattered sources into a consistent foundation that can be queried and explained.

  • PythonProcessing & validationFrom two files to a review queue.
  • pandasPreparation & reconciliationA comparison starts by defining what is comparable.
  • SQLQueries & business rulesInvestigate what is missing from the breakdown.
  • BigQueryMarketing data modelsWhen a number does not add up, I trace its path.
  • Google CloudData infrastructureUnderstand how data reaches the report.

02 / 07 · From workflow to application

Automation & development

Connect interfaces, APIs and automation to turn a data workflow into a useful application.

  • ReactInterfaces & interactive experiencesMake an analysis explorable.
  • JavaScriptInteraction & application logicEvery change on screen needs a clear rule.
  • HTML & CSSStructure & responsive designGive complex information a clear hierarchy.
  • FastAPIData services with PythonSeparate the analysis from the screen displaying it.
  • FlaskApplications for data workflowsBring data collection into an application.
  • openpyxlExcel file processingRead the workbook before assuming its structure.

03 / 07 · Analysis people can explore

BI & reporting

Present metrics so others can explore them and make decisions.

  • StreamlitAnalysis & QA interfacesQA that people can work through, not just run.
  • Power BIDashboards & performance analysisGive different signals a shared interpretation.
  • Data StudioMarketing reportsA report starts with the question it must answer.
  • Microsoft ExcelData analysis & reviewFrom a received file to a defensible review.

04 / 07 · From signal to outcome

Measurement & CRM

Connect digital behavior with commercial context and outcomes.

  • Google Analytics 4Behavior & conversionsConnect the arrival with what happens next.
  • SalesforceLead & business contextUnderstand what each lead stage represents.

05 / 07 · My work with paid media

Media & AdTech

Understand spend, delivery and outcomes across advertising channels and platforms.

  • Google AdsCampaign analysisUnderstand what explains the result.
  • Search Ads 360Paid searchA KPI needs a shared definition.
  • Campaign Manager 360Campaign measurementCheck what we are counting.
  • Display & Video 360Programmatic display & videoConnect the detail to the total.
  • Meta AdsAudiences & performanceThe same campaign, in both sources.
  • TikTok AdsVideo & creative performanceSeparate attention from response.
  • Reddit AdsAudiences & interestsFind where the difference begins.
  • Microsoft AdsSearch & intentCheck coverage before comparing.
  • LinkedIn AdsProfessional audiencesOrganize the assessment by objective.

06 / 07 · Brand, conversation & context

Listening & intelligence

Understand conversations, trends and demand to bring context to brand decisions.

  • BrandwatchConversations & perceptionDistinguish presence from perception.
  • EmplifiSocial performanceFrom social data to a performance assessment.
  • MeltwaterMonitoring & reputationConnect monitoring with brand analysis.
  • TalkwalkerConversations & brand signalsAsk what lies behind a conversation spike.
  • SemrushSEO & demand signalsCompare what people discuss with what they search for.

07 / 07 · Build, check, iterate

Applied AI

Support exploration, development and communication while retaining human judgment and review.

  • OpenAI CodexBuilding & reviewing codeIterate on an application that already solves a problem.
  • ChatGPTAnalysis, ideas & communicationTurn a model idea into explicit rules.
  • OpenAIReporting automationConnect analysis with an explanation people can review.
  • CopilotSupport for daily workReach a useful first draft sooner.
  • HiggsfieldVisual exploration & videoExplore how an idea can be told visually.
  • Hugging FaceModels & prototypesExplore a model from the problem outward.
  • AntigravityApplication developmentGive the agent a task that can be checked.
  • ClaudeAnalysis, code & documentationPut the explanation through another review.

Contact

You can message me on LinkedIn to discuss marketing analytics, automation or applied AI.

Message me on LinkedIn