Urde et al. (2007) describe a heritage brand as one that makes its history part of its present identity, and they identify five elements by which the strength of that heritage can be judged: a track record, longevity, core values, the use of symbols, and a belief within the organisation that its history matters. Heritage of this kind is hard for competitors to copy. It also carries an expectation of continuity. Beverland (2005), writing about wine, shows that authenticity has to be actively maintained through visible commitment to tradition and craft, and that it is damaged when commercial motives become too apparent. Work on retro brands adds that a brand's past can be made attractive to people who never lived it, provided the story is retold and not merely repeated (Brown et al., 2003). The past, in other words, can speak to a young urban audience, but only if the brand has not discarded it.

Research on extensions explains the risk to the existing base. New products that contradict what buyers believe about a brand can weaken those beliefs for the parent (Loken and John, 1993). Kirmani et al. (1999) found that current owners of a brand respond differently from non-owners when its line is stretched upward or downward in price, and that sub-branding can protect the parent when a stretch would otherwise be resented. Against this caution stands the evidence on how brands grow. Sharp (2010) argues that growth comes mainly from reaching more buyers, most of them light and occasional, and that the customer profiles of competing brands differ less than managers suppose. On that view a tea brand should hesitate before treating rural and urban buyers as separate worlds, and should above all avoid damaging the memory cues and the distribution on which every sale depends.

Writers on rebranding reach a compatible conclusion. Muzellec and Lambkin (2006) place rebranding on a continuum from evolutionary to revolutionary change, and Merrilees and Miller (2008) propose that a successful revision keeps enough of the core to form a bridge from the old brand to the new, while meeting the needs of new segments and carrying internal stakeholders along. The framework below is an attempt to turn that advice into a sequence of decisions.

The first task is to find out what is actually held in memory. For rural buyers this means qualitative work in the places where tea is bought and made: which features identify the pack, what the brand is trusted for, what would make a buyer suspect a change. Laddering interviews, which move from product attributes to their consequences and then to personal values, are suited to the purpose (Reynolds and Gutman, 1988). The same exercise is run with urban millennials. The planner also lists the brand's distinctive assets, the colours, shapes and phrases that trigger recognition, and how widely each is known (Romaniuk and Sharp, 2016). The safeguard at this stage is one of order: the rural study is commissioned first and reported first, so that its findings frame the brief for everything that follows.

The audit is then used to divide the brand's elements into those that are fixed and those that may be adapted. Table 1 offers a starting classification for a tea brand. It is a template for the audit to correct, not a finding. The governing rule is that anything rural buyers use to recognise or trust the brand is fixed unless there is strong reason otherwise, and that the heritage story is treated as common property of both audiences, to be told differently to each but never contradicted.

With the core fixed, the planner decides how the urban offer will be named. Aaker and Joachimsthaler (2000) describe a spectrum of options running from a single master brand, through sub-brands and endorsed brands, to wholly separate brands. For a heritage tea the two ends are unattractive. Changing the master brand itself puts the whole base at risk, and a separate brand throws away the heritage that is the main advantage with younger buyers. The framework therefore recommends a sub-brand or an endorsed range as the default: new formats carry the parent's name and symbol together with a descriptor of their own. The safeguard is that the base pack is left exactly as it is while the new range is introduced. Any modernisation of the base pack is treated as a separate, later project.

The choice between the two forms depends on what the audit shows. Where urban non-buyers respect the parent name, a sub-brand that leads with it is appropriate. Where the name itself is the obstacle, an endorsed range gives prominence to a name of its own and shows the parent as guarantor, which keeps some distance while still drawing on the heritage. The ownership findings of Kirmani et al. (1999) suggest that existing buyers of an everyday brand tend to accept an upward stretch, which is a further reason to prefer a visible link to concealment.

The urban range is then designed as a complete offer: product, pack, price, outlets and communication. Trade-offs between attributes such as format, origin claim and price can be explored with conjoint analysis before anything is produced (Green and Srinivasan, 1990). Separation of channels does much of the protective work. If the new range is sold where rural buyers seldom shop and advertised in media they seldom see, the two audiences rarely meet each other's version of the brand. Separation is never complete, since families span village and city and social media travel, so the urban communication should be checked for anything that a rural buyer would read as disrespect for the original. The safeguard is commercial: the price, pack sizes and rural distribution of the base product are held constant through the launch. Affordability in small units is central to low-income markets (Prahalad, 2005), and a base that sees its sachet price rise as a premium range appears will draw the obvious inference.

The launch is staged, beginning with a limited number of cities, and monitored in both audiences. Among urban buyers the planner tracks awareness, trial and repeat purchase of the new range and whether it changes perceptions of the parent. Among rural buyers the measures are defensive: base volume by district, retailer stocking, and periodic checks on trust and on any belief that the product has changed. The safeguard is a set of stop rules agreed before launch. These state which movements in the rural measures would pause the roll-out, who decides, and what corrective action follows. Thresholds agreed in advance are harder to argue away than those invented under pressure.