Social media has always rewarded noise. But when that noise is engineered — liked, shared and commented on by thousands of phones in a warehouse — it undermines the very metrics media professionals rely on.
Recent reporting from Fast Company suggests bot activity has hit a new threshold: more organised, more subtle and much harder to detect. It’s no longer just a political problem, but a commercial one.
Today’s bot farms aren’t run by crude scripts. They’re physical operations: racks of real smartphones, programmed to pass as people. Enough of these devices like or share a post and the algorithm reads it as momentum and pushes content further, regardless of origin.
In theory, this should be the moment agencies and brands start questioning what they’re optimising for. In practice, many still treat trending content, spike graphs and sentiment heatmaps as clean signals. But if amplification can be bought for pennies, and the platforms aren’t filtering it out, then what exactly is everyone measuring?
The problem with performance
Performance is still the primary language of strategy, even as its meaning deteriorates. Campaigns that ‘work’ are often campaigns that move quickly, not necessarily ones that resonate with people. In a system optimised for velocity, relevance can be faked. So can traction — and yet the habits around measurement persist.
‘Brands have over-relied on vanity metrics for too long. Bot amplification just makes the cracks impossible to ignore,’ says Callum McCahon, executive strategy director at Born Social. ‘Social signals like likes and shares are easy to inflate, and even easier to misread. At Born Social, we see this as a necessary wake-up call and a reminder that engagement was never the goal.’
McCahon sees the current landscape as a moment for recalibration: a shift from chasing visibility to building cultural weight. ‘The brands that will win now,’ he says, ‘are those that understand real brand-building happens beneath the surface — through fame, cultural relevance, emotional resonance, and ultimately, commercial impact.’
It’s a persuasive line, but also a revealing one. If engagement was never the goal, why did so much strategy — and so many tools — grow up around it?
The answer, perhaps, is convenience. Likes and shares are easy to track, easy to sell, and easy to wrap into campaign decks. Metrics tied to cultural resonance or long-term brand impact are harder to isolate and slower to surface. They’re also harder to fake, which may be precisely why they matter now.
‘Success demands a shift in measurement,’ McCahon adds, ‘moving beyond engagement rates to metrics that prove impact on business outcomes.’ But measurement systems don’t change on insight alone. As long as reporting frameworks and internal KPIs orbit engagement, strategy remains shaped by what’s visible, not necessarily what’s valuable.
What lies beneath the dashboard
Some agencies have started asking better questions. Paul Greenwood, global head of research and insight at We Are Social, says they now steer clients toward engagement types that are ‘much harder to fake: saves, UGC creation, repeat comments — not just raw likes.’
It’s a subtle reordering of priorities: away from visibility and toward signals that suggest intent. Saves take effort, UGC implies trust and comments (the real kind) signal community more than virality.
‘Sentiment analysis has always been verified at We Are Social by pairing it with human review/coding and deeper behavioural signals,’ Greenwood says. ‘Vetting influencer audiences and monitoring for inorganic spikes is standard.’
While it’s a methodical approach, even these practices assume the data they start with is reliable. Greenwood notes that manipulation is ‘fairly easy to spot if you’ve been working in social for long enough.’ That may be true if you know what to look for. On social platforms, engagement is abstracted and context is stripped away. For many media planners and brand teams operating inside of them, often at speed, the distinction between what’s real or synthetic is not always clear.
The gap between internal rigour and external conditions is the real challenge. You can build better metrics, but if the raw inputs are corrupted, your interpretation — no matter how sophisticated — still risks being led astray.
The comment section knows
Influencer marketing positions itself as the more human edge of digital media: community-driven, creator-led, less reliant on blunt metrics. But it’s no less exposed to manipulation. The difference is that the scrutiny often starts closer to the ground.
‘Our relationship with influencers and their management is the most effective way of identifying inauthentic activity online — it’s all about the human touch,’ says Hannah Ryan, head of campaigns (UK/US) at The Goat Agency. ‘Yes, there are tools that can spot fake followers and engagement, but when we review talent we look at their recent engagement rates and analyse why they might have received higher engagement or an outlier on a particular post. If we see something that looks unexpected, we’ll review and determine thoroughly if this is genuine engagement or not.’
This suggests a form of reading the room that doesn’t show up in spreadsheets. Ryan places particular weight on comment sentiment. ‘If you see genuine conversation below a post, then the followers are going to be genuine. If it’s just loads of emojis, spam or one-word comments, then they’re likely bots. Comments underneath posts reflect how audiences engage with the authentic content they see.’
‘Fake followers used to be able to hide a few years ago,’ she adds, ‘but the market is much more mature now and it’s easier to spot and flag. It takes a combined approach from agencies, influencers and their management to target inauthentic interactions. Open communication and dialogue help keep insights accurate for advertisers while protecting influencers from bad players online.’
What Ryan’s describing isn’t a data fix, but a cultural shift. One where instinct, collaboration and shared knowledge play a bigger role in verifying what performance actually represents. It’s not that influencer marketing is immune to distortion. It’s that, in many cases, it’s already adapted to working in a contaminated environment.
From measurement to meaning
2024 made it harder to ignore how far upstream the distortion runs. In the US, bot-led manipulation muddied political sentiment using the same mechanics that now underpin social strategy. In Russia, bot farms used TikTok to push emotionally charged propaganda about the war in Ukraine, gaming the same systems brands rely on to track reach and resonance.
The platforms have done very little since to change how any of this works. The signals are still noisy.
There’s a shared undercurrent in everything these agencies are saying: not that measurement is broken, but that it can’t be assumed to mean what it used to. Signals need context, spikes need scrutiny and even performance needs a second look.
Bot-led amplification is ambient now. It shapes the same metrics used to validate creative, benchmark influencers and report back to clients. The tools built to detect it at both platform and strategy level aren’t moving fast enough.
So the job has changed. The brief is no longer to chase what worked. It’s to understand what made it work, and whether it should have. Because when amplification is cheap and authenticity is optional, media strategy becomes an exercise in echo-reading. And the longer you steer by echo, the further you drift from signal.