Guides
The three things Dutch agencies keep getting sold hype about.
Attribution, tool consolidation, and AVG-compliant tracking, each explained the honest way, with no fabricated numbers. Written to be useful before you ever talk to us.
Guide 01
Server-side attribution: why platform numbers stopped matching client revenue
18 min read · Updated June 2026
For years the number in the ad platform and the number in the client's accounting told roughly the same story. That reconciliation was the real product an agency sold: proof that the media worked. It is now the thing most quietly breaking, and it is breaking for reasons outside any single campaign.
This guide explains what changed, why it costs agencies retainers, and the specific mechanism that fixes it: a first-party pixel that keeps the click ID, identity resolution that ties an anonymous click to a named person, and server-side conversion upload that sends the truth back to the platforms.
What actually broke
Three shifts happened at once. Apple's App Tracking Transparency let users opt out of the tracking that Meta relied on. Safari's Intelligent Tracking Prevention capped and cleared the cookies that browser pixels depend on. And the third-party cookie, the mechanism that let a platform recognise a visitor across sites, is being retired across the board.
The result is that a growing share of conversions never reach the ad platform in the browser at all. Platforms fill the gap with modeled and estimated conversions. The number on the screen still moves, but it drifts away from the number the client can actually find in their revenue.
Why the gap costs retainers
When the dashboard stops reconciling to real sales, every review meeting turns into a debate about whether the numbers are real. The agency ends up defending its measurement instead of demonstrating its results, and that is exactly the moment a client starts questioning the retainer.
The honest fix is not a better dashboard on top of the same broken data. It is to change where and how the conversion is measured, so the report is built from deals that actually closed rather than browser events the platform guessed at.
The fix, part one: a first-party pixel
A per-tenant first-party tracking snippet runs on a domain you control and sends events to your own server, authenticated per tenant. Because it is first-party and server-processed, it survives the browser-side blocking that cripples third-party pixels.
Critically, it captures the ad click identifiers at the source and preserves them: fbclid, gclid, wbraid, gbraid, li_fat_id, ttclid and the rest, plus full UTM and referrer data. That preserved click ID is the raw material for everything downstream.
The fix, part two: identity resolution
A click ID is useless until it is tied to a person. On identify or lead capture, the visitor resolves to one canonical person record, the captured click IDs are folded onto that person, and the earlier anonymous session is linked back.
This is the bridge that lets a sale be attributed to the specific ad click that started it, even after a long, multi-channel, multi-week journey. An fbclid that arrived three weeks before the form fill is now attached to a real, named lead, and later to a won deal.
The fix, part three: server-side conversion upload
When a deal advances a stage, for example a call is booked or a deal is won, Metiva fires a server-to-server conversion to the ad platforms with hashed data and the original click ID. Real adapters exist for Meta CAPI, Google (Enhanced Conversions for Leads plus Offline Conversion Import), LinkedIn CAPI, and TikTok Events.
This is engineered to be reliable rather than a fragile script. A deterministic event ID is used as both the dedup key and the log key so retries are idempotent, there is a per-deal advisory lock, backoff retries honour the vendor's Retry-After, and an hourly monitor tracks Meta Event Match Quality.
- Conversions can be mapped per deal stage to per-platform events, with sensible defaults seeded automatically.
- A tracking-health view shows per-platform success, failure, and retry counts, plus match quality and last-event age.
What you get back
The ad platforms are now told about the conversions that actually became revenue, server to server, so their optimisation aims at real downstream sales instead of cheap form fills. And because the pipeline and the attribution live in the same system, the report you hand the client and the revenue they can see are finally the same story.
Guide 02
How to consolidate your agency tool stack without breaking delivery
14 min read · Updated June 2026
The typical Dutch agency runs close to ten tools, each with its own monthly invoice, and none of them share a record. Consolidation is worth doing, but not the way it usually gets sold, as a heroic weekend migration. It is a phased move that proves the new system on live work before switching off the old one.
This guide is honest about where a point-tool stack genuinely wins, what the real cost of fragmentation is, and the order in which to consolidate so delivery never breaks.
The stack you are actually running
Named honestly, the stack a Dutch agency pays for today looks like this. It is a lot of tools for one connected job.
- A shared inbox (Trengo) for client messages.
- A CRM and pipeline (Teamleader, Pipedrive).
- Projects and work tracking (Simplicate, Gripp).
- Visitor identification and de-anonymization (Leadinfo, Albacross).
- Client reporting (Swydo).
- Invoicing (WeFact).
The real cost is not the subscriptions
The subscriptions are the small number. The large number is the reconciliation tax: the hours spent copying a deal into the projects tool, matching the reporting tool against the invoicing tool, and answering which lead came from which campaign when the answer is spread across four systems.
The same client exists ten times across ten tools and never quite matches. Every mismatch is manual work, and every piece of manual work is a place for the numbers to drift.
Where a point-tool stack genuinely wins
This is the honest part. A best-in-class point tool is often deeper in its one job than any consolidated system, and if a single function is your whole differentiator, the specialist may be the right call. A stack also lets you swap one tool without touching the others.
The trade-off is that you own the integration work forever, and the one thing a stack structurally cannot do is tie an ad click to a closed deal, because the click and the deal live in different databases.
The order of operations
Consolidate in phases, starting with the surfaces that share the most data and cause the most double entry. Prove each phase on real work before cancelling the old subscription.
- Phase one, the spine: the unified inbox and the lead-to-deal CRM, so every message and deal resolves to one person record.
- Phase two, delivery and money: contracts, e-signature, invoicing, and payments, plus the client portal, so a signed contract converts straight into an invoice.
- Phase three, the measurement layer: first-party tracking and server-side attribution wired to the deals you just consolidated.
What not to consolidate
Do not rebuild accounting. Metiva is built to connect to Moneybird and Exact Online, not to replace them (those adapters are on the roadmap, framed honestly). Consolidation is about ending the reconciliation tax, not owning every last function. Keep what already works and does not need to share the person record.
Guide 03
AVG-compliant tracking: consent-first measurement for Dutch agencies
12 min read · Updated June 2026
For a Dutch agency answerable to the AVG, the Dutch implementation of the GDPR, compliant tracking is not a disclaimer to bury in a footer. It is a reason to buy. The agencies that get this wrong either over-collect and carry real risk, or under-collect and lose the ability to prove the media worked.
This guide covers the posture that lets you do both: measure what proves ROI, and stay squarely inside the AVG.
Deny-by-default consent
The foundation is that targeting forwards to the ad platforms are suppressed unless consent has been decided and granted, rather than firing first and asking later. That single default is the difference between a setup you can defend and one you have to apologise for.
It pairs with Google Consent Mode v2, the framework for adjusting how tags behave based on a visitor's consent choices, deny-by-default in the EU.
IP truncation, encryption, and erasure
IP addresses are truncated and encrypted at rest, with full-IP reveal audited rather than routine, and configurable storage and retention modes so you keep only what you need for as long as you need it. In production the system hard-fails without an encryption key configured, so there is no accidental plaintext path.
Right-to-erasure is built in, so a delete-visitor-data request is a supported operation rather than a scramble.
Company-level versus person-level
This is the distinction that trips people up. When a visitor has not consented, Metiva can still surface the company or account behind a visit, which is legitimate business context. It resolves to a named individual only when that person consents through a form capture.
Said plainly: person-level tracking is the consented case, not the default. The honest version of the de-anonymization claim is company-level identification where consent is not present, person-level only on consent.
Why this still proves ROI
You do not need to track everyone to prove the media worked. The conversions that matter, the ones that become revenue, come from consented, identified leads who filled in a form and became deals. Track the right people correctly, let the company-level layer cover the rest, and the attribution still reconciles.
Start here
Want this mapped to your actual stack?
The Tool-Stack & Attributie Audit is a 5-day audit into a 0 to 365-day rollout plan. We map your current tools and your real attribution gap, in Dutch or English.
No slides. We reply within one business day. Your data stays in the EU.