Growth & Search
One team owning organic, AI search and paid media — on top of tracking, tagging and page performance we build ourselves. For companies tired of three vendors reporting three different numbers.
Most companies do not buy digital marketing services. They buy three retainers. SEO from one vendor, paid media from another, content from a third, and an internal analytics person trying to reconcile all of it in a spreadsheet on Friday afternoon. Each vendor optimises the metric they are graded on, each reports a version of reality that flatters their own channel, and nobody owns the number the board actually asks about. ZenMagix is a software engineering agency in Mumbai. We came to marketing from the systems side — we spent years building exchange infrastructure and payment rails where a number was either right or the money was gone — and that is the posture we bring here. Attribution, event tracking, data pipelines and page speed are engineering problems. Most marketing agencies subcontract them or ignore them. We treat them as the foundation, then run channels on top, and we tell you plainly which parts of your reporting are measured and which parts are modelled.
When SEO, paid and content sit with different vendors, they optimise separate metrics that conflict. The damage is not dramatic — it is a slow tax you never see itemised.
Here is what it looks like in practice. Your paid team buys brand terms because it makes their blended CPA look excellent, and a large share of those clicks would have arrived free through organic. Your SEO vendor publishes forty top-of-funnel guides because sessions are their KPI, and none of them map to a product your sales team can close. Your content agency writes for a keyword list nobody validated against closed-won deals. Your ads land on a page nobody optimised for speed, so a third of mobile visitors leave before the hero image paints, and the paid team reports it as a creative problem.
Each vendor is doing their job. The system is still losing money. Nobody is accountable for the interaction effects because no contract mentions them. Coordination is not a soft benefit here — it is where most of the recoverable waste sits. We run the channels as one programme against one set of numbers, with a single funnel model everybody is measured on. When paid and organic compete for the same query, we decide deliberately rather than discovering it in a quarterly review. When a landing page is the constraint, we fix the page instead of testing another headline.
Attribution is broken and getting worse. Roughly a third to half of what most dashboards report as fact is modelled, and the honest version of marketing reporting says which is which on every line.
Four things degraded at once. Browser privacy controls and ITP shortened cookie lifetimes, so multi-week B2B journeys fall apart before conversion. Consent requirements mean a meaningful share of sessions is never observed at all, and platforms fill the gap with modelled conversions presented in the same font as measured ones. Dark social — a link forwarded in WhatsApp, dropped in a Slack channel, mentioned on a call — arrives as direct traffic with no referrer and no story.
And AI search referrals are the newest hole: a buyer reads about you inside ChatGPT or an AI Overview, forms an opinion, and turns up later as a branded search that your SEO report happily claims. Every platform is incentivised to over-claim, which is why the sum of channel-reported conversions routinely exceeds the number of deals you actually closed. We do not pretend to solve this. We reduce the unknown where engineering can — server-side tagging, first-party identifiers, offline conversion imports back into ad platforms, CRM-joined reporting — and then we label the rest.
Measured, modelled, or inferred, on every number. On top of that we run geo holdout tests and periodic spend pauses, because for the questions that actually matter, a controlled experiment beats an attribution model.
Correct event tracking, server-side tagging, clean data pipelines and fast pages decide whether any of the above works. We build them in-house, in your repository, because we are an engineering company that also does marketing.
Ask a marketing agency to deploy server-side GTM on Cloud Run, or to fix an LCP failure caused by an unoptimised hero and a render-blocking font, or to build a dbt model joining GA4 exports to CRM stages, and the answer is usually a referral to a partner or a ticket filed with your developers that sits for two months. That gap is where most marketing programmes fail quietly. A tracking plan that was never enforced means your conversion events fire twice on some templates and not at all on others, and every downstream decision inherits the error.
We start engagements by auditing the data layer before we touch a campaign, because optimising toward broken events is worse than not optimising at all. Then we build: a documented event schema, server-side tagging with consent handling wired in properly, conversion APIs into Meta and Google so platform optimisation sees real outcomes, GA4 and Search Console exports landing in BigQuery, dbt transformations producing tables a director can read, and Core Web Vitals treated as field data at the 75th percentile rather than a lab score screenshot. It ships as pull requests your engineers review. That is the whole differentiator, and it is not a positioning exercise — it is just what a software company can do that a media buyer cannot.
This page is the frame. The actual work lives in three specialised practices — organic search, AI search visibility and paid media — coordinated under one plan rather than sold as one undifferentiated bundle.
Organic search is the compounding asset and the slowest to move: crawlability, rendering, site architecture, structured data and content built against bottom-of-funnel intent. AI search visibility is the newer surface — being retrieved and cited inside AI Overviews, ChatGPT and Perplexity, which depends on the same foundations plus passage-level clarity and entity consistency, and which almost nobody is measuring properly yet. Performance marketing is the fast lever: paid search, paid social and the creative iteration loop that decides most of the outcome.
We sequence them rather than launching everything in month one. Usually that means fixing measurement and page performance first, buying traffic while organic compounds, and shifting the mix as the numbers justify it. Now the honest part. This is not for companies who want a Rs-per-month retainer and a slide deck. It is not for anyone who needs leads next Tuesday with no tracking in place, wants a guaranteed cost per lead before we have seen the funnel, or expects an agency to compensate for a product with no distinct position in its market. If the brief is cheap and monthly, there are hundreds of agencies in Mumbai who will take it, and some of them are fine. We are a poor fit and we will say so on the first call.
What you get
A full review of your data layer, tag configuration, conversion events and platform-reported numbers against your CRM. You get a written account of what is genuinely measured, what is modelled, where double-counting is inflating channel credit, and what it costs to fix each gap.
Documented event schema, server-side tagging with consent handling, Conversions API integrations, GA4 and ads data landed in BigQuery, and dbt models joining marketing activity to CRM stages. Deployed as pull requests into your codebase and owned by you afterwards.
One plan covering organic, AI search and paid, with stated assumptions, a budget split, what we expect each channel to produce by when, and explicit notes on which forecasts are confident and which are guesses. Revised against actuals every quarter, not quietly forgotten.
The running work: paid search and social campaigns, creative iteration cycles, landing pages built and optimised in-house, and content produced against briefs derived from your closed-won deals rather than a volume-sorted keyword export.
A single dashboard where every channel is measured against qualified opportunities and revenue, each figure labelled by confidence, alongside incrementality test results. Plus a monthly written read on what changed and what we got wrong, which is usually the more useful half.
How we work
Three to four weeks auditing tracking, tag setup, historical channel performance, landing page field data and CRM records. We reconcile what platforms claim against deals you actually closed, and the gap between those two numbers usually sets the agenda.
Rebuild the event layer, deploy server-side tagging, wire offline conversions back into ad platforms, and fix the Core Web Vitals failures costing you paid and organic traffic. Nothing scales on top of numbers you cannot trust.
Launch in sequence against a shared funnel model — usually paid for immediate signal while organic and AI search visibility compound. Creative and content iterate on a fixed cadence, with decisions logged so you can audit why the mix changed.
Geo holdouts and structured spend pauses to test incrementality, quarterly reallocation based on what the tests show, and reporting that separates measured from modelled. We tell you when a channel has stopped earning its budget, including ours.
Every phase ends at a decision point you can stop at — see how that works across fixed-scope projects, embedded pods and retainers.
Stack
Questions
We are a software engineering company that runs marketing, so tracking, server-side tagging, data pipelines and page performance are built in-house rather than subcontracted or filed as tickets with your developers. That means the measurement layer under every channel is something we own and can fix, not a dependency we complain about.
No percentage of spend — it rewards us for spending more, which is a bad incentive to build into a contract. Engagements start with a fixed-fee diagnostic and infrastructure phase, then move to a monthly retainer scoped to engineering and channel hours, invoiced against a defined scope. Media budget is paid directly by you to the platforms.
Sometimes, and it works best when we own measurement and infrastructure while they keep executing their channel. What does not work is two vendors both claiming credit from separate dashboards with no shared funnel model. If the incumbent will not report against a common set of numbers, the arrangement usually fails within two quarters.
Because nobody honestly can. Cookie lifetimes are short, consent gaps hide sessions, dark social arrives as direct traffic, and AI search referrals often carry no referrer at all. We narrow the unknown with server-side tracking, CRM joins and self-reported attribution, then use holdout tests for the questions attribution genuinely cannot answer.
Paid channels give signal within weeks once tracking is correct, though the first month is usually spent finding out how wrong the previous numbers were. Organic and AI search visibility compound over six to twelve months. The measurement and performance work delivers value immediately, because it stops you optimising toward events that were firing incorrectly.
Both, though the work differs. B2B means long cycles, low-volume high-intent terms and CRM-joined reporting on qualified pipeline. D2C means creative volume, contribution margin after returns and shipping, and marginal ROAS rather than blended. The engineering layer underneath — clean events, server-side tagging, fast pages — is identical.
Questions about cost, timelines, IP ownership and data residency are answered on the general FAQ, and how this practice came out of blockchain infrastructure explains why we build the way we do.
Related
Paid acquisition run against contribution margin, not vanity ROAS.
Getting cited by ChatGPT, Perplexity and AI Overviews — measured, not promised.
Engineering-led SEO: crawl budget, rendering, structured data, Core Web Vitals.
See also: build the compounding organic foundation with engineering-led SEO services · get retrieved and cited in AI answers with AI search visibility · run paid search and social against real pipeline with performance marketing
Tell us what you are trying to build. We will tell you honestly whether we are the right team for it, and what it would realistically take.
