Which engagement metrics do clients actually value? The ones you can connect to a business decision. Likes and reach show that something happened; saves, shares, comments with intent, branded search and incremental revenue show that it mattered. A report that stops at the first list is window dressing. One that gets to the second list persuades.
A case published this week shows the way. According to Campaign Middle East, performance agency TRKKN used Meridian, Google's open-source marketing mix model, to show beauty retailer Faces where its revenue really came from. The modelling covered three years of data across 11 channels and projected an 8.5% incremental revenue lift and a 10.8-percentage-point improvement in ROI, purely by reallocating budget that already existed.
Harry Lonergan, digital marketing director at Faces, summed up the gain: Meridian helped the brand look beyond what last-click performance can show.
Why don't likes convince clients any more?
Because clients don't pay for likes. They pay for sales, for brand, for market share. When the monthly report opens with "engagement is up" and doesn't say what that changed, the marketing director asks the question every agency dreads: "So what?"
The problem isn't the engagement metric itself. It's treating it as an end rather than a signal. Engagement is the symptom; the client wants the diagnosis.
What is the data illusion effect?
It's the tendency to mistake numbers for understanding. A dashboard full of percentages looks objective, precise, scientific. So we stop asking whether the metric actually measures what matters. The precision of the number disguises the weakness of the question.
With engagement, the illusion shows up in three ways. Comparing rates across platforms that define engagement differently. Celebrating volume without reading what people actually said. And crediting the whole result to the last click, the very trap that, according to Lonergan, Meridian helped Faces escape.
Data reveals patterns, but it cannot define humanity. A high rate could be a passionate fan base or a pile-on. Only reading the content tells you which.
Engagement metrics: window dressing versus what persuades
| Metric | What it shows | When it matters to the client |
|---|---|---|
| Reach and impressions | How many people saw the content, or how many times it appeared | As context for the other metrics, never on their own |
| Likes | A quick, low-effort reaction | Almost never, in isolation |
| Engagement rate | Interaction relative to reach | When compared within the same platform and the same format |
| Comments | What people think, in their own words | When they are read and classified by theme and intent |
| Saves and shares | Usefulness, and a willingness to attach one's own image to the content | Almost always: they are strong signals of value |
| Branded search | Interest that has left the feed and turned into intent | To prove brand and campaign effect |
| Incremental revenue and ROI | What the media actually added to the business | In the budget conversation, with marketing and finance |
How do you compare engagement across different platforms?
Carefully and, ideally, without adding the numbers up. Imagine an agency running the social channels for a women's football club. Instagram shows a healthy engagement rate on behind-the-scenes content; TikTok, lots of views on match highlights; YouTube, watch time on full matches. Lumping it all into one "total engagement" figure erases precisely what each number is telling you.
The better route is to compare each platform with itself over time and, for the overall picture, to look at the people. Who are the fans commenting across all three channels? What tensions come up, such as struggling to find out where matches are being shown, or wanting to get to know the players better? That you can compare across platforms, because it's content, not a proprietary metric.
In niche sports, that kind of reading is worth its weight in gold. The audience may be smaller, and precisely because of that, one well-read comment says more than a thousand impressions.
How do you build an engagement report the client actually reads?
- Start with the client's decision. Renewing a sponsorship, launching a product, defending the budget in next year's plan. The report exists to serve that decision.
- Pick three metrics per objective, at most. Any more and it becomes an inventory, not an analysis.
- Connect each attention metric to a business one. Saves to site visits; comments to branded search; reach to regional sales.
- Read the content, not just the number. Classify comments by theme and intent. Three real quotes are worth more than a chart.
- Hold forecasts up against results. If you said a creative would work, show whether it did. That is how credibility is built.
- End with a recommendation. Where to add, cut or hold budget in the next cycle. A report without a recommendation is just the minutes of a meeting.
Step five is where a lot of agencies lose credibility without noticing. On the Defy platform, predictions are validated against real ad performance, and every conclusion can be checked against the posts behind it. So the results conversation with the client starts from evidence, not from a defence.
When is it worth moving to marketing mix modelling?
When the client's question stops being "did the post do well?" and becomes "where should the next chunk of budget go?". The TRKKN case shows the potential: with Meridian, Faces identified channels with room to grow, channels close to saturation and budget that could be moved elsewhere. Performance Max emerged as the best-performing channel.
But MMM needs history. The project used three years of data, online and offline, across 11 channels. For smaller clients or newer accounts, start with the basics done well: the right metrics, qualitative reading and simple incrementality tests. The goal is the same either way: to swap the illusion of the number for a real understanding of what moves the business.
