The cold-start problem describes a situation where a system cannot function well until it has accumulated data, but it cannot accumulate data until it is already functioning. In market intelligence, the analogue is the information asymmetry facing a deal team that is entering a sector it has not covered before.
Established investors in a sector know the names. They know which companies are founder-controlled and which are PE-backed. They know which management teams have expressed interest in a transaction and which are firmly not engaged. They know who the peripheral players are that rarely show up in any market map. This knowledge was built over years of conversations, sector conference attendance, and informal contact. A new entrant to the sector has none of it.
Why the incumbents' advantage is structural, not just positional
It is tempting to frame the cold-start problem as a positional disadvantage that effort can overcome: if you make enough calls, attend enough events, and read enough trade press, you can assemble the universe over time. The reality is that the incumbent advantage compounds faster than any realistic cold-start effort can close the gap.
Every transaction that an established investor completes in a sector generates new information. The due diligence process surfaces the names of competitors that the target's management team considers credible. The banker running the sale process names the other bidders who received the information memorandum. Post-close, portfolio company management teams become informal intelligence networks, referring potential targets and flagging sector developments. None of this information enters the public record, and none of it is accessible to an entrant that has not yet done a deal in the space.
The public record can not replicate this. What it can do is provide the structural foundation: the universe of companies that operate in the sector, their approximate sizes, their ownership status, and their organisational signals. That foundation is not a substitute for relationship-based intelligence, but it is the prerequisite for building it. You cannot prioritise relationship-building if you do not know which relationships to prioritise.
The three phases of cold-start market entry
In practice, a deal team entering a new sector goes through three phases. The first is universe construction: identifying all the companies that could plausibly be in scope for a mandate. The second is prioritisation: ranking those companies by acquisition readiness indicators. The third is engagement: making contact and building the relationship intelligence that converts a structural map into a live pipeline.
The cold-start problem is sharpest in phase one. A team without prior sector coverage has no list of names to start from. The obvious approaches produce incomplete universes. SIC code searches retrieve too many irrelevant companies and miss relevant ones. Keyword searches on company descriptions miss companies that do not describe themselves in the expected terms. Trade association membership lists are partial and often out of date. Industry publications reference the most prominent companies and systematically under-represent the mid-market tier where most acquisition activity actually takes place.
The resulting universe is not empty; it is biased. It systematically over-represents companies that are visible in public sources and under-represents those that are not. In the UK mid-market, many of the most interesting acquisition targets are companies that have never issued a press release, never attended an industry event, and have no web presence beyond a basic Companies House filing. They are invisible in any search-based approach and visible only in the filing record.
What the filing record contributes to universe construction
A structured approach to the filing record changes the universe construction problem from one of recall (finding all the relevant companies) to one of filtering (removing the irrelevant ones from a complete set). The starting point is the full population of UK-registered companies in the relevant SIC code range. That population is large and heterogeneous, but it is complete. Every company that operates in that sector and is registered in the UK is in it.
Filtering that population down to acquisition-relevant companies uses three types of signal. Size indicators narrow the field to companies within a mandate's revenue range. Activity indicators confirm that the company is actually operating in the target sector and not just categorised there by administrative default. Ownership indicators establish whether the company is independently owned or already part of a group that would make an acquisition either impossible or unnecessary.
This approach does not eliminate the cold-start disadvantage entirely. The filing record tells you that a company exists, operates at a certain scale, and is independently owned. It does not tell you whether management is open to a conversation, whether the company has had a prior process, or whether there is a competitive dynamic that would make an approach awkward. Those layers require the relationship intelligence that only comes with time in the sector.
What it does is compress the timeline of the first phase significantly. Universe construction that previously required two to three weeks of analyst calls and desk research can be completed in a fraction of that time, leaving more capacity for the activities that are genuinely relationship-dependent.
The case for structural intelligence as the baseline
There is a reasonable argument that relationship-based sourcing should come first and structural analysis second: if you know the right people, you will hear about the best opportunities before they reach any formal process, and structural analysis is mainly useful for identifying companies you have not already heard about through your network.
That argument holds reasonably well for established sector specialists. It does not hold for teams entering new sectors, and it does not hold for any team operating across multiple sectors simultaneously. At some point, the relationship network has finite bandwidth, and the mandate set is wider than any individual network can reliably cover.
The more fundamental issue is that a relationship-first approach systematically under-covers the parts of any sector that are least networked. The most relationship-rich companies in any sector are the ones that have interacted with the financial community before, either through a prior process, a banking relationship, or management background that includes investment banking or private equity. Companies that lack this background are consistently under-reached by relationship-first sourcing, not because they are less attractive, but because they are harder to find through relationship channels.
The cold-start problem is in part a cold-start problem for the market, not just for a specific deal team. Many of the best targets in any sector have never had a structured conversation with a financial buyer, not because they are not interesting, but because no one has found them through the relationship network. A structural approach that starts from the full population of the market and filters down, rather than starting from the relationship network and extending outward, is more likely to surface these companies. That is the asymmetry that makes the filing-based approach worth building.