Should a Brand Sound the Same Every Time? Brand Response Consistency and Customer Evaluations
On social media, a brand's replies are public and easy to compare. When a brand answers every customer in a similar way, does it look reliable or robotic? Our research shows the answer depends on the customer's mood.
Brand replies are no longer private
Customers increasingly turn to social media instead of calling customer service. A brand's reply on X is public, it stays online, and anyone can compare it with the replies the brand gave to other customers. Brands also manage these replies more and more through response guidelines, templates and, now, generative AI.
Most research asks whether a single reply is good. We ask a different question: how does a reply look next to the brand's other replies? We call this brand response consistency: how much a reply matches the brand's usual way of responding, in both meaning and wording.
Two sides of consistency
Consistency sends two signals at once. On one hand, a reply that looks like many others can feel less personal. It suggests the brand did not really pay attention to this customer. On the other hand, consistency signals that the brand handles customers in a stable, fair and predictable way. Which signal matters more depends on the situation.
What we studied
We analyzed 359,117 brand replies and 80,546 customer follow-up messages across 207 brands on X. We looked at two audiences: the customer who wrote to the brand, and other users who saw the exchange, measured by their likes.
What we found
• On average, more consistency means less favorable reactions. Customers' follow-up messages are less positive, and observers give fewer likes.
• The customer's mood changes the picture. The more negative the customer's first message, the weaker this penalty becomes.
• For very upset customers, consistency helps. When a customer is strongly negative, a consistent reply is linked to a more favorable follow-up. These customers seem to value a reliable, fair process over a personal touch.
• Observers react the same way, but less strongly. Observers dislike consistent replies most when the customer's message is positive. For negative messages, the penalty shrinks but does not fully reverse.
Consistency is a dial you can turn
Generative AI makes it easy to write a new, unique reply every time. So should brands stop caring about consistency? Our results say no. We show that firms can control how consistent their AI-written replies are:
• Prompt design: giving the model example replies (few-shot prompting) makes replies more consistent.
• Temperature: higher temperature settings generally make replies less consistent.
• Model choice: different models produce different levels of consistency.
The takeaway
Consistency is neither good nor bad. It is a setting to match to the tone of the conversation. With happy or neutral customers, a fresh, personal reply works better. With angry customers, a reliable, consistent reply can work better. Brands should also treat their replies as one public portfolio, not as separate conversations, because other customers are watching.