Somewhere in my first serious attempt to write a methodology section, I got stuck on a question that sounds almost embarrassingly basic: is Brand Drift a real thing, or is it just a story I've decided to tell about a pattern I noticed? That question sat with me for weeks longer than it should have, mostly because the two obvious answers both felt wrong in a way I couldn't yet articulate, which is its own uncomfortable place to sit as a researcher trying to write a defensible chapter.
The two easy answers, and why I couldn't take either one
The first easy answer is pure positivism: treat Brand Drift as something you can only claim exists if you can directly measure and observe it happening — survey data, sentiment scores, a clean causal chain from cause to effect. The appeal is obvious; it feels rigorous. The problem is that it quietly assumes reality is nothing more than whatever gets captured in your measurement, which is a strange thing to assume about a phenomenon whose defining feature is that it operates beneath anyone's notice until it's already advanced. If drift is only real when someone's measuring for it, then by definition it can't be the slow, unnoticed thing I keep finding in every case I study — the compound-interest error nobody flagged for months, the chatbot policy nobody was watching until a tribunal got involved. Positivism, taken strictly, can't even describe the phenomenon I'm trying to study, because the phenomenon's whole character is that it happens in the gap between what's occurring and what anyone's actually observing.
The second easy answer swings the other way entirely: treat brand identity as purely discursive, a story constructed differently by every observer, with no underlying structure to speak of — just competing narratives, none more accurate than another. That's tempting too, especially once you've read enough postmodern brand theory to be suspicious of any claim to objective truth. But it has its own dead end. If brand identity is nothing but whatever story each observer tells, then "drift" can't mean anything either, because there'd be no stable thing to drift away from in the first place — just an endless series of equally valid interpretations, none of which could ever be called mistaken.
I needed a stance that could hold both halves at once: something real enough to actually drift, and something knowable enough that my access to it, however partial, could still be meaningfully right or wrong — rather than a stance that quietly picked one half and pretended the other half didn't need answering.
What Bhaskar's stratified reality actually gives you
Critical realism, the philosophical position Roy Bhaskar developed beginning with A Realist Theory of Science, splits reality into three layered domains instead of collapsing it into one. There's the Real — the underlying structures, mechanisms, and causal powers that exist and operate whether or not anyone is observing them, the way a fault line exists whether or not an earthquake has happened yet. There's the Actual — the events those mechanisms actually produce when triggered, whether anyone happens to be watching or not. And there's the Empirical — the narrow slice of actual events that someone actually experiences, records, or notices.
This maps onto Brand Drift with a precision that genuinely surprised me the first time I laid it out. The mechanism — a production system scaling faster than the oversight built to govern it — sits at the level of the Real. It's a structural condition, present in an organization whether or not anyone's currently measuring it. The drift itself, the actual gap widening between stated identity and produced output, sits at the level of the Actual — happening in real time, whether or not a journalist, a tribunal, or an internal auditor has noticed it yet. And what gets written up as a "scandal" — CNET's correction notice, Air Canada's tribunal ruling — sits at the level of the Empirical: the small, delayed, imperfect slice of the actual drift that someone finally happened to catch and document.
That's the piece pure positivism can't hold: the Real and the Actual keep operating whether or not the Empirical catches up to them, which is precisely why so many of these cases run for months before anyone notices. And it's the piece pure discourse analysis can't hold either: the mechanism producing drift isn't just a story someone's telling about it — it has a structure that would keep producing the same effect in a different company, in a different country, whether or not anyone had a name for it yet.
Why the honesty about partial access matters just as much
Critical realism doesn't let me off the hook by granting me the Real domain and calling it a day, though. It pairs that realist claim with something called epistemic relativism — the honest acknowledgment that any given researcher's access to that real, structural layer is always partial, shaped by which cases they happened to study, which theories they brought to the table, and which language their sources were written in. I take that seriously enough that it shapes the actual design of my research: four countries, not one, specifically because my access through a single national case would be too partial to trust on its own. Estonia earns its place in the design precisely as the check on my own partiality — the case chosen to see whether my account of the mechanism survives contact with a context I didn't build the framework from in the first place.
What critical realism refuses to do, though, is collapse that honesty about partial access into "so nothing can be judged more accurate than anything else." It holds onto what's called judgmental rationality — the position that even though every account is partial, some accounts still explain more of what's actually happening than others, and that's a real, defensible distinction worth making, not just a matter of preference. That combination — real structure, honestly partial access, and the confidence to still say one explanation is better than another — is, as far as I can tell, the only intellectually honest way to study something that's simultaneously structural and quiet.
What this looks like when I'm actually building an interview protocol
The abstract version of this is easy enough to nod along with. The harder part was translating it into an actual research instrument. When I interview a founder about how AI got introduced into their brand's customer touchpoints, I'm not treating their account as a direct, unmediated window onto the Real mechanism — that would quietly slide back into naive positivism, as if their perception simply were the structure itself. But I'm also not treating their account as just one more story with no privileged claim on what actually happened — that would slide into the discourse-only position I already rejected. I treat it as empirical evidence, generated by someone with genuine but partial and situated access to an actual event, which is itself a partial manifestation of a deeper structural mechanism. Every interview is doing inference across all three layers at once, which is slower and more uncomfortable than either easy alternative, and which is exactly why I trust the findings more when they hold up under that heavier weight, rather than when they simply feel intuitively right on a first pass.
Why this isn't just throat-clearing before the "real" research
I used to think the methodology chapter was the part you get through so you can reach the interesting findings. I don't think that anymore. The methodological stance is the reason the findings are trustworthy at all — it's the difference between "I noticed something and I'm calling it a pattern" and "here's a structural mechanism, here's how I know my access to it is partial, and here's why I still think this account explains more than the alternatives." Only one of those survives being handed to someone who disagrees with you, and only one of them, honestly, deserves to survive that scrutiny in the first place.
There's a version of this that sounds like unnecessary philosophical throat-clearing until you've actually tried to defend a finding without it, at which point it stops sounding academic and starts sounding like the only thing standing between a genuine contribution and a plausible-sounding anecdote. I'd rather spend the extra months getting the stance right than publish something faster that collapses the first time someone asks me the question I asked myself for weeks before I had an answer worth keeping.
If you're working through your own research design and stuck on whether your stance actually fits your phenomenon, that's a conversation worth having early, not after the data's collected. Get Your Free Brand Audit →
References
Bhaskar, R. (1975). A Realist Theory of Science. Leeds Books.
Bhaskar, R. (1979). The Possibility of Naturalism: A Philosophical Critique of the Contemporary Human Sciences. Harvester Press.
Fletcher, A. J. (2017). Applying critical realism in qualitative research: Methodology meets method. International Journal of Social Research Methodology, 20(2), 181–194.