Smarter Data Inputs for Better Algorithm Performance with Anni Salo

In the modern digital marketing landscape, everyone is obsessed with collecting as much data as possible. But are we focusing on the right kind of data?

In this episode of The Profit and Proof Podcast, host Joie Roberts speaks with Anni Salo of Tracklution about why marketing data quality matters more than simply collecting more data. They discuss how better conversion signals can improve advertising algorithm performance, why marketers need a reliable source of truth, and how AI and large language models are changing search, content creation, and marketing measurement.

Here is a breakdown of the core insights from their conversation.

The Myth of “More Data”

Marketers today are hoarders. We collect massive amounts of data, yet many teams have no idea where it actually comes from or how accurate it is. Anni points out that the industry is suffering from a data quality problem, not a quantity problem.

Algorithms—whether on Meta, Google, TikTok, or Snapchat—need the right “fuel” to work properly. If you feed an algorithm messy, inaccurate, or fragmented data, you cannot expect it to generate profitable results.

Surviving the “Black Box” Algorithms

Year after year, major advertising platforms have been stripping away the manual controls that marketers used to rely on to optimize their campaigns. Today, these platforms operate as a “black box”—you don’t fully know what happens inside; you just see the clicks and conversions that come out.

With manual tweaking no longer an option, Anni explains that the only real lever marketers have left is controlling the data inputs. By feeding the machine higher-quality signals, you train the algorithm to sift through the noise and find your ideal customers.

However, this requires a Single Source of Truth. Platforms like Meta and Amazon naturally over-attribute sales to themselves because they want you to increase your ad spend. To avoid flying blind, brands must stop relying solely on platform-reported numbers and establish a unified, accurate tracking system on their own backend.

How a Tracking Crisis Shaped Tracklution

Anni didn’t just learn this theory in a textbook; she lived it. A few years ago, while running a performance marketing agency, Apple rolled out a major iOS privacy update. Almost overnight, her agency lost tracking visibility and about half of their conversions. Revenue dried up, salaries were at risk, and the company faced bankruptcy.

Instead of giving up, her CTO spent six grueling months building a custom data engine to bypass the tracking issues. They used it purely to save their own business- and not only did they recover, but they doubled their revenue.

When industry friends started asking how they survived the iOS update, Anni realized that smaller brands and marketers didn’t have the IT resources or developers to build their own tracking tools. They decided to pivot the entire company to share this technology with the world, eventually naming it Tracklution.

AI, LLMs, and the Future of SEO

The conversation also tackled the rapid rise of Large Language Models (LLMs) and how they are shifting the internet from “search” to “answer-based discovery.”

Anni notes that we are currently experiencing a blast from the past—it feels like the early days of SEO and Google Ads. Because every user prompts AI differently, it is nearly impossible to predict exact keywords. Instead, marketers must go back to the basics: creating genuinely useful, human-centric, high-quality content.

Because AI allows us to generate content faster than ever, the internet is flooding with noise. To stand out and get picked up by LLMs, your content (whether it’s a podcast, video, or blog) must offer real human experiences, emotions, and undeniable quality.

Anni’s Advice for Adopting AI

For businesses feeling overwhelmed by the pace of AI, Anni offers a reassuring perspective: LLMs are at their worst today—they will only get better tomorrow.

Her top advice for marketers?

1. Don’t wait for a course: The technology is moving so fast that by the time a course is published, it’s outdated. You have to jump in and learn by doing.

2. Solve a specific bottleneck: Find one daily task or problem that slows you down, and test different AI tools to see if they can automate it.

3. Keep the human element: Let AI handle the heavy lifting, but leave the most important, empathetic tasks to the human brain.

Anni’s advice is simple: Start with a real problem, experiment with new AI tools, and learn through practical use. Whether the tool solves the problem or teaches you something new, the exercise creates value.

Want to audit your own tracking setup? Connect with Anni Salo on LinkedIn or visit Track.com to try out Tracklution’s AI tracking assistant.