Nine hundred images, posted by six people over fourteen months. Not one of them had ever changed a decision we made.
That channel was proof of diligence, and diligence is not the same as learning.
Collection feels like work
The reason competitive research rots is that the collecting part is pleasant and the thinking part is not.
Screenshotting a rival’s new ad takes nine seconds, produces a visible artefact, and earns a small social reward when someone reacts to it. Working out what that ad implies about their positioning, whether it represents a test or a commitment, and what you should therefore do differently, takes forty minutes and produces nothing anyone can see.
So teams do the nine-second version repeatedly and call it competitive intelligence. We certainly did. The channel grew, the diligence was visible, and our actual strategy was unaffected by any of it.
What a screenshot cannot tell you
Here is the specific failure. A single image is a sample of one, with no context and no time dimension.
You cannot tell from it whether the ad is new or has been running for eight months, which is the single most useful signal available, because longevity is the closest thing to a public performance metric any of us get. You cannot tell whether it represents a direction or an experiment, because one ad is not a pattern. And you cannot compare it against the rest of their library, because the rest of their library is not in your channel.
What you have is a thing that looks like evidence and functions as an anecdote. And anecdotes are extremely good at confirming whatever the loudest person in the room already believed, which is how we spent a quarter convinced a rival had “gone all in on UGC” on the basis of four screenshots.
Structure is the whole difference
What changed it for us was not more collection. It was the arrival of structure underneath the collection.
Using Hawky’s competitor analysis, every competitor ad comes in tagged by hook archetype, visual hierarchy, message framing, CTA and offer, across Meta and Google. Because the tagging is consistent, the library becomes something you can ask questions of rather than scroll: which hooks is this brand testing most often, how has their creative shifted over ninety days, what angle has nobody in the category touched.
The shift is from an image to a row in a dataset. One screenshot is an anecdote. Four hundred consistently tagged ads are a distribution, and a distribution supports an inference.
The second thing that mattered was time. Seeing their testing cycles and refresh cadence over months tells you what they are committing to versus merely trying, which is exactly the distinction a screenshot destroys.
The side effect I did not anticipate was what it did to internal arguments. Our competitive conversations used to be settled by whoever had looked most recently, because nobody could check anyone else’s claim. Once the library was shared and structured, a disagreement became something two people could resolve by filtering, in about a minute, rather than by taking turns asserting. That changed the tone of planning meetings more than it changed any individual decision.
Where I still do not trust it
Tagging is interpretation, and interpretation can be wrong. A hook classified as “problem-agitate” might be read quite differently by the person who wrote it. The tags are good enough to find patterns and not good enough to settle an argument about intent, and I have seen people treat them as the latter.
It also tells you what they are doing and not why. A competitor may be running a weak offer because their margin changed, because someone senior insisted, or because they are clearing stock. The creative looks the same in all three cases, and your counter-move should not be.
And there is a trap in finding the open angle. The page frames an untested angle as an opportunity, and sometimes it is. Sometimes it is untested because four companies already tried it quietly and it did not work. Absence of evidence is doing a lot of work in that framing, and I have burned a month on a gap that turned out to be a hole for a reason.
Closing Thoughts
The uncomfortable thing I took from counting those screenshots is that our competitive research had been a ritual rather than a practice. It made us feel informed and left us unable to answer any question a CFO might ask.
The test is simple and most teams fail it. Point at a decision you made in the last quarter, and trace it back to something you learned about a competitor. Not something you noticed. Something that changed what you did.
If you cannot find one, you do not have a competitive intelligence problem. You have nine hundred screenshots.