Bidding on value, not volume
Each stage carries a value, so Google and Meta bid on revenue instead of lead count. Value-based bidding only works when those values arrive reliably.
Connect ad spend to actual revenue
Growth analytics for banks, fintech and e-commerce. I build the measurement that cuts wasted ad spend and finds the growth last-click reports miss.
Deduplicated and consented
Audit report in two weeks
- Website
- Mobile app
- CRM
- Call tracking
- Offline sales
One identityServer-side GTM, CAPI and BigQuery- Google Ads
- Meta CAPI
- TikTok Events API
- AppsFlyer S2S
Every source resolved to one identity, modelled once, then pushed back to the platforms that spend the money.
Selected clients
What changes when the platforms see the whole funnel
Most accounts send a form fill from the browser and nothing after it. The algorithm then finds more people who fill forms, not people who pay.
Send the outcome, not just the lead
Qualified leads, approvals and sales go back to Google Ads and Meta from your CRM through the Conversions API and offline conversion imports.
The platform sees 24 identical leads and looks for more people like all of them.
- Became a customer
- Junk or duplicate
- Still in the funnel
The interfaces above are mockups with sample numbers.
Start with a diagnosis. Not a contract.
Four services. Most engagements start with the audit and move up only if it finds something worth building.
- Attribution auditFind out which numbers you can trust.$3,000, 2 weeks
- Server-side GTMServer-side GTM, built to survive ad blockers.Scoped, 4-8 weeks
- GA4 setupGA4 that answers business questions.Scoped, 4-8 weeks
- Mobile attributionMobile attribution that survives iOS.Scoped, 4-8 weeks
The audit, in two weeks
- 1Intake call
- 2Read-only access
- 3Funnel traced end to end
- 4Ranked report and roadmap
Better signals start before the form
Illustrative interface with sample data. No live account is connected.
A quiz before the form
A short quiz sorts visitors before the form. Only qualifying answers are sent as conversions, so campaigns stop optimising for curiosity clicks.
Leads checked against government data
On submit, the phone and IIN are checked against the mobile citizens database and the business against the tax registry. Only verified leads become conversions.
Recent cases
What the work looks like, by type of client. The numbers for each case are in the portfolio.
Server-side events
An auto-lending MFO
Browser and server sent the same registration twice with no shared ID. GTM now sends one event ID, and lead forms reach the CRM through a signed webhook.
In the portfolioQualified leads only
A leasing company in a bank group
Every form submit counted as a lead. The form now checks the tax ID first, and GTM marks a lead qualified only after it passes.
In the portfolioOne event taxonomy
An insurer inside a super app
Purchase events did not match signed contracts. A shared taxonomy is live on one product; rollout to the rest is in acceptance.
In the portfolioTracking without code access
A consumer lender
The client could not change the site, so an interceptor reads the funnel's own API calls and records every step.
In the portfolioSEO on a single-page app
An e-pharmacy marketplace
Moved the storefront to server-side rendering and fixed canonicals and sitemaps, so search engines could index the catalog.
In the portfolioClean production analytics
A large Kazakh bank
Found test traffic and phantom users in production analytics, cut the tag manager down and redesigned the app taxonomy.
In the portfolioPut those signals to work
Illustrative interface with sample data. No live account is connected.
Real results for every creative
CRM statuses land next to each ad in the account, including lead forms and messenger campaigns that usually stay a black box.
Predicted value from day one
A pLTV model scores each sign-up from early signals, and the predicted value reaches the ad platforms long before the first payment.
Campaigns learn faster
The platform gets the whole deal history: application, qualification, payment and rejection. It drops spam sooner and brings leads that pay.
Three ways to work together
The audit and the retainer have fixed prices. A project is priced once its scope is agreed.
Attribution audit
Start here$3,000fixed scope, 2 weeks
A read-only review of how your funnel is measured, and of the gap between what the platforms report and what your CRM settles.
- Findings ranked P0 to P3, each with root cause and fix
- Quick wins split from structural work
- A 90-day roadmap
- The report is yours either way
Project
Scoped4-8 weeks
One build, agreed before it starts: server-side GTM and CAPI, an MMP or S2S pipeline, an event taxonomy, or an offline conversion import.
- The working pipeline in your own accounts
- The spec that documents it
- A handover your team can maintain
- Fixed scope, milestone updates
Fractional retainer
$2,500per month
Standing analytics engineering capacity. I own the measurement layer: new tracking as the product ships, QA when a platform changes the rules, and the reporting on top.
- No long-term contract, pause anytime
- Unlimited analytics requests
- Average 48-hour turnaround
- Direct line on Telegram or Slack
Keep the whole journey in view
Illustrative interface with sample data. No live account is connected.
GA4 sees what happens after the form
Approvals and payments happen in your backend. Measurement Protocol sends them to GA4, so funnel reports reach revenue.
One stream, every platform
The browser sends one first-party request. Your server-side GTM container decides what each platform receives and strips personal data before it leaves.
Counted once, not twice
Browser and server send the same purchase with one shared ID, so the platform keeps one and drops the copy.
Products from my own work
GitHub profileTools I built for client work and then packaged. Each comes with setup help if you want it.
Marketing MCP connector
Claude gets your GA4, Google Ads, Search Console, Tag Manager, BigQuery and Yandex accounts through one login, deployed on your server.
How it worksMarketing agents for Claude Code
Nine specialist skills plus reviewer and legal agents, and a commit gate that waits for review.
Public release soonAudit toolkit
Open-source audits for Google Ads, GA4, Search Console and Meta, plus semantic diffs for Tag Manager.
How it workspulsecheck
A Chrome DevTools recorder for tracking audits: network, dataLayer and consent, exported to Markdown.
How it worksTracking plan from Figma
Generates the event taxonomy from your design files and flags drift when the designs change.
How it worksAcademia
A free course on digital marketing and analytics on goodlabs.kz.
Open AcademiaIllustrative interface with sample data. No live account is connected.
Product capabilities reviewed on 13 October 2026 against the project README.
Hi, I'm Dilshat Rakhimov
Digital marketing leadership, with the technical depth to build what comes next.
Good Labs is my consulting practice for banks, insurers and lenders. Alongside my role at inDrive, I work on attribution, conversion infrastructure, product analytics and acquisition.
I connect business goals to technical requirements, then stay involved through implementation and testing. That perspective comes from leading digital marketing at Halyk Bank and Bereke Bank, as well as working directly on the systems behind acquisition.
Based in Almaty, with consulting experience across Kazakhstan, Russia, the Czech Republic and India.
How the work actually gets built
All field notesTechnical writeups from client engagements: architecture, event schemas, and the failure modes that cost the most to find.
- Marketing mix modeling: what it measures, how it works and when it pays offAdstock and saturation, ROI vs marginal ROI, and the data you need before you start.
- Conversion Lift: the holdout test that shows which conversions your ads causedIn the worked example the platform attributes 3,000 conversions and the experiment finds 900.
- Geo experiments with GeoX and GeoLiftHoldback, go-dark and heavy-up, power and MDE, and calibrating MMM with the result.
Let's talk about your project
I start with a two-week audit. Fixed scope, no long-term commitment upfront. If it surfaces value, we continue. If not, you keep the roadmap.
dilshat@goodlabs.kzt.me/arcbaslowAlmaty, UTC+5, remote-first