WhatcompetitorreportshavetaughtusaboutusingAI

At Branagh PR, we produce monthly competitor analysis reports to help clients understand how other businesses in their market are communicating and where that leaves their own approach. Each report follows an agreed group of competitors, examining their website activity and media coverage alongside wider industry developments. We bring that research together with our analysis and recommendations, giving management and marketing teams a clearer basis for deciding whether, and how, to respond.

Every month, I put these reports together for management and marketing teams who want to understand what is happening in their market and whether it should influence their own plans. Over time, the work has also become something of an education in using AI, particularly in recognising how easily a well-presented finding can feel more reliable than it is.

There is no question that AI has broadened the research. I can cover more ground and uncover stories that a manual search might miss, with more time left to consider what they mean. Yet a finding arriving neatly summarised, complete with a source link, can give the impression that some of that thinking has already been done. It is tempting to accept the explanation and move on, especially when there is plenty more material waiting.

Finding more does not always mean knowing more

Our process starts with a powerful AI tool that monitors changes across an agreed group of competitor websites and links back to the pages behind its findings. We generally follow the same companies each month, allowing us to build a picture of their activity over time, although that group can change as new rivals become relevant to a client.

Alongside this, we use AI tools such as Copilot and Claude to help search for third-party media coverage, with Google Alerts providing another route into stories. Wider industry news is important context here too, because a competitor’s announcement takes on a different significance when you understand the conditions surrounding it. A change in messaging may be more revealing if several businesses are beginning to respond to the same concern.

As the material accumulates, patterns start to emerge. An individual award win might warrant a mention, but repeated recognition over several months could suggest that a competitor is building a reputation the client needs to take seriously. Equally, a newly worded product page may look like a change in direction until a closer read shows that very little has changed. Having more information gives us more to work with, but it also gives us more to weigh up.

This is where we slow the process down

Before a finding reaches the report, I follow it back to its source, check that it falls within the reporting period and read what the page actually says. The AI summary can help me find my way into a story, but I still need to establish whether its interpretation holds up.

I have seen searches return material from outside the dates requested, as well as summaries that misread what an article says about a competitor. Sometimes the problem is less obvious than an incorrect fact: the wording may sound certain, although the source leaves room for doubt. That difference matters when a client may use the report to inform a decision.

Some of the time AI saves therefore goes straight back into checking its work, which seems to me a reasonable exchange. The broader search is valuable precisely because we can examine what it brings back. Without that step, we risk passing on outdated information or giving an assumption more weight than the evidence supports.

Even a fully checked finding needs another kind of scrutiny before it becomes useful advice. Knowing the account changes how we read the research, because a campaign that creates pressure for one client may point to an opportunity for another. We have to consider what the business is trying to achieve and whether a response would serve that purpose.

The recommendations are where the report earns its value

Our reports have developed with this in mind. The early versions were produced in document format. Our current, more visual format helps readers compare competitors side by side and follow the research through to our recommendations. Looking across their activity and media presence makes it easier to see where a client may be losing ground, or where there is room to contribute something competitors have overlooked.

Content gaps often become clearer at this stage. If rivals repeatedly discuss an issue the client has barely addressed, we consider whether there is a useful perspective missing from the client’s communications. That does not automatically mean publishing something similar. The stronger response might be to answer a question others have avoided, drawing on knowledge the business already has.

Those recommendations come from us, informed by the research and our understanding of the client. A finding might lead to a media angle or prompt us to revisit website messaging, but it needs a reason beyond the fact that a competitor has done something.

The hardest page may be the shortest

Perhaps the most demanding part is then bringing the work down to a one-page executive summary. With less room to explain, I have to be more decisive about what deserves the reader’s attention. Something can be interesting enough to include in the full report without being significant enough to lead the summary, and making that distinction takes thought.

For someone who cannot read every section immediately, this page should offer a clear starting point, with the fuller evidence there when they need it. Writing it brings me back to the question running through the whole process: what does this client need to take from what we have found?

AI has made that question better informed, although the answer still depends on conversations and context beyond the search results. I might know that a suggested campaign would stretch a small team too far, or that an apparently modest change is important because of a discussion we had with the client months ago. When I am deciding what belongs on that final page, those details are key. They are often what turns a finding into a recommendation I would be comfortable talking through with the client.