Trackingplan VS InvoiceIn

Trackingplan's answer

Trackingplan is used by brands and agencies including dentsu, Havas Media Network, Schneider Electric, El Corte Inglés, Cofidis, RIU Hotels & Resorts, Euronics, ISDIN, Baleària, Wpromote, Making Science and Awaze.

Trackingplan's answer

In 2021, our founders faced a familiar frustration: broken analytics and unreliable data. Dashboards didn’t add up, key events were missing, and making decisions felt like guesswork.

Determined to fix it, we built Trackingplan; a tool to automatically monitor, validate data tracking, and catch issues before they impact business decisions. What began as a solution for ourselves quickly became essential for companies around the world.

Since then, we have been building the best observability layer for analytics data, sitting exactly where data breaks: the hit, the dataLayer, the SDK, the pixel, the CAPI payload, the consent state, the GTM release.

That’s where the truth of a number is actually decided, before it becomes the number everyone relies on.

But observability was only the foundation. Agency is the destination.

With AI, the question is no longer whether a system can act on your data. It’s whether the signal it acts on can be trusted.

Agents can now fix a tag, pause a campaign, rewrite an attribution model, or move your budget and bids. And the more autonomous they become, the more their decisions depend on the quality of the signal underneath them.

That’s where Trackingplan comes in: the observed state of your tracking, checked against what should be there, with a clear answer when reality and expectation don’t match. Because most AIs are trained to answer, ours is trained to tell the truth.

InvoiceIn's answer:

Every EU e-invoicing mandate is written from the sender's side: issue in a structured format, deliver over a network. The receiving side got far less attention — even though that is where the obligation lands first. German businesses have had to be able to receive since January 2025; French ones since September 2026.

So a company that must accept e-invoices finds two kinds of tooling: Java libraries it has to host and keep current, or full networks it has to join. Nothing in between for the plain question — I have this file, what does it say, and is it valid?

InvoiceIn is that in-between piece, built by a small independent studio that ships narrow, oddly specific products.

Trackingplan's answer

Trackingplan is the only digital analytics agent that works where data is produced, not where it ends up. Instead of reading a warehouse or a dashboard after the fact, it observes every request a website, iOS and Android app, and server sends to 80+ analytics and ad platforms, along with the dataLayer, consent signals and tag manager releases behind them. That's where a number becomes true or false.

This gives the agent something general-purpose AI doesn't have: evidence. It learns each company's events, properties, and normal traffic patterns from live data, with no data model or tracking plan to set up. So it can tell broken tracking apart from a real change in the business, trace an issue to its root cause, and show the exact hits behind every answer. Teams ask questions in plain English, get reports and audits written for them, and receive results in Slack, Teams, email, Jira, or Claude.

InvoiceIn's answer:

It only does the receiving half — and does every European format through one endpoint.

Most e-invoicing products are networks: they sign you up, connect you to Peppol, and send on your behalf. InvoiceIn does none of that. It takes a file that already arrived — over Peppol, by e-mail, by portal download, by a customer upload — and turns it into data you can use.

• One endpoint accepts XRechnung, ZUGFeRD/Factur-X, Peppol BIS 3, EN 16931 UBL, CII D16B, FatturaPA and KSeF FA(2)/FA(3), as XML or as a hybrid PDF.
• The output shape is the same whatever came in: canonical EN 16931 JSON, fields named by their BT/BG number.
• Every validation finding carries a plain-language fix in five languages, not just a rule id.
• Nothing is stored. The file is parsed in memory and discarded with the response.

Trackingplan's answer

Against general-purpose AI (ChatGPT, Claude, Gemini connected to a warehouse): an AI can only reason about what it can see. A warehouse shows what eventually arrived. It doesn't show whether a pixel fired correctly this morning, whether consent changed, or whether a GTM release dropped a parameter. Generic AI fills those gaps with a plausible guess. Trackingplan answers from first-party observations of every hit, so it knows when the data is wrong and says so, with the evidence.

Against AI assistants built into analytics platforms (GA4, Adobe, Amplitude): each one only sees its own platform and assumes its own data is correct. Trackingplan sees every destination at once, so it can explain why Meta reports more purchases than GA4, or why one platform is missing a property the others receive.

Against traditional tracking QA and monitoring tools (ObservePoint, Avo and others): those tools depend on predefined crawls, test scripts or a tracking plan maintained by hand. Trackingplan learns the implementation from real traffic, monitors it 24/7 across web, apps, and server-side, and goes beyond alerts: it investigates, explains, writes reports, and runs audits on a schedule.

It's also fast to adopt: one tag or SDK, no data model to build, a 14-day free trial with no credit card, and pricing based on traffic rather than seats.

InvoiceIn's answer:

Against the open-source libraries — Mustang, the KoSIT validator, the Factur-X toolkits: they are excellent and free, and they are Java or Python packages you host, update, and keep in sync with quarterly rule-set releases. If your stack is PHP, Node or Ruby on shared hosting, that is not a small ask. InvoiceIn is an HTTP call.

Against the platforms — Basware, Avalara, Tradeshift: they solve the whole procure-to-pay problem, with onboarding, contracts and seat pricing. If all you need is to read the invoices you already receive, that is a large purchase for a small job.

InvoiceIn sits in between: no infrastructure to run, no platform to join. Pay per document, credits never expire, and the first 20 a day need no key at all.

Trackingplan's answer

Teams that depend on digital analytics and marketing data being right, at mid-market and enterprise companies with websites and mobile apps:

• Digital analytics teams, who need to know when a release breaks tracking before it breaks reporting.
• Paid media and performance marketers, who need conversions, UTMs and Meta CAPI or Google Ads signals to be accurate before bidding algorithms optimize on them.
• Compliance, legal and DPO teams, who need to catch PII leaks, consent breaches and undeclared cookies on every hit.
• Agencies and consultancies that manage analytics for many clients and want an always-on analyst for every account without growing headcount.

Common industries include ecommerce and retail, travel and hospitality, financial services, consumer brands, and media and marketing agencies.

InvoiceIn's answer:

Two groups, arriving for different reasons.

Product teams at accounting, AP and ERP vendors, e-commerce and shop-plugin developers, and agencies building tooling for SMBs. Their users started receiving e-invoices and now have to read them — and writing a parser per format, then maintaining it, is not the product they set out to build.

In-house finance and AP teams in Germany, France, Italy and Poland, where receiving is already mandatory or becomes so in 2026–27, and supplier files arrive in whatever format the supplier chose.

InvoiceIn's answer:

• Python + FastAPI for the API.
• lxml for XML parsing and XSD validation.
• The official Schematron rule sets, compiled to XSLT 2.0 and executed on SaxonC — the same artefacts the KoSIT validator uses, so a finding here matches a finding there.
• pikepdf to pull the embedded XML out of ZUGFeRD and Factur-X hybrid PDFs.
• WeasyPrint for the human-readable PDF rendering.
• Deployed as a Docker container on a European server, behind Cloudflare.