A sourcing strategy is a deliberate decision about which places outside your own website a company must have a presence on, so that systems form a correct and complete picture of it. It is based on the fact that generative search engines do not answer from a single source but assemble an answer from several, and evaluate discrepancies between them as uncertainty. A company that has an excellent website and is nowhere else is, from their point of view, poorly substantiated. The first step is to find out where information on the given topic is drawn from. For the set of questions that matter to you, you record which domains appear as sources in the answers. The result tends to be a surprise: it is often specialist portals, government websites, encyclopaedias, industry directories and discussion forums that the company never thought of. This list is the basis of the entire strategy – it tells you where the game is being played. The second step is dividing the sources by reachability. Some of them you can influence directly, for example by entering and maintaining data in catalogues, registers and professional networks. Others you reach through work – an expert contribution, a comment for the media, providing your own data to a journalist, a case study published by a partner. And some you cannot reach at all, which is also useful information, because it saves budget. The third step is aligning the data. At all reachable places, the same basic information must appear: the name in a consistent form, address, contact details, description of activity, link to the website. Discrepancies are worse than missing entries, since they signal to the system that it does not know which piece of data is valid. The most common source of discrepancies is old profiles left over after a change of registered office or name. The fourth step is building mentions in places that carry weight. It is not about quantity but about thematic proximity and the credibility of the source. One mention in a specialist overview that models cite on the given topic does more than twenty entries in catalogues with no audience. A mention need not contain a link – even the mere association of the company name with the topic and with an expert context strengthens the entity. The fifth step is dealing with content that is not yours and that describes you inaccurately. The solution is not to complain about the model but to fix the source: updating an outdated profile, adding the correct data to a catalogue, a factual reply in a discussion, or possibly your own page that clearly resolves the given question and gives the model something to cite. The sixth step is regular review. The composition of sources changes over time, and so does where it is worth having a presence. Measurement should therefore include not only whether you are mentioned, but also the list of cited domains – a change in it tends to be the first signal that the topic has opened up to a new player or, conversely, closed off. Of the whole of GEO, the sourcing strategy is the part that is hardest to buy and slowest to build, yet also the hardest to imitate. When planning the work, it is worth dividing sources into three waves according to the ratio of benefit to effort. The first wave is entries and profiles that you fully control and whose alignment is a matter of hours – registers, maps, main catalogues, professional networks, social media profiles. The second wave is industry directories, associations and partner websites, where someone needs to be contacted, but the barrier is low. The third wave is media and specialist portals, where content or data worth publishing is needed. Most companies start with the third wave, because it is the most visible, and never finish the first – even though it is precisely the one with the best ratio of result to time spent. A useful side effect of the whole strategy is that the same data alignment also helps in classic search and in local results, so the investment does not pay off in just one channel alone.
See also: Source consensus, Digital PR for AI Visibility, Source Diversity in AI Answers.
