From Admission to Outcome: The Conversion Chain
An incubator’s effect unfolds as a sequence, and the subject here is what happens at each transition between an offer of support and a startup result. The practical question is where the chain can break: before a team participates, while support is being delivered, when readiness changes, or when that change has to travel into customers, financing, survival, or growth. Following the sequence makes it possible to diagnose an apparently simple failure—“the programme did not accelerate us”—without assuming that every link was ever in place.
The path from an offer to a useful dose
The first transition is from selection to participation. An incubator may scout or select promising young companies, and selection can be valuable if it places a startup among relevant peers or gives it access to scarce support. But admission is not the same as an intervention. A selected company may have looked promising before the programme began. Its later performance can reflect its team, technology, market, or prior relationships rather than anything the incubator did.
This is the central reason to separate selection from treatment. If an incubator admits companies that are already more capable or better connected than non-participants, a comparison between members and non-members will tend to credit the programme for differences that existed beforehand. The success of admitted firms may be real while the causal contribution of the incubator remains unknown. To learn what support changed, one must ask what was offered, to whom, compared with what alternative, and what changed after the offer.
The second transition is from an offer to delivery. A programme description can promise coaching, specialist advice, testing, peer learning, workspace, or introductions. Delivery means that the promised resource is actually available at the time and quality the startup needs. A mentor assigned on paper may have no relevant expertise or too little time. A testing facility may be technically impressive but inaccessible within the startup’s schedule. An investor event may take place without producing conversations relevant to the companies attending. The service exists in the programme’s inventory, but not necessarily in the startup’s working environment.
The Western Balkans investment-readiness programme illustrates why the dose matters. Its intensive offer was not a single workshop. Firms were given an online platform for examining their problem, commercialisation concept, revenue streams, market size, and competitive position; lead-mentor feedback; individual mentoring from a larger pool of specialists; masterclasses; and pitch preparation. The comparison group received a much cheaper online course. The contrast therefore contained different amounts and forms of work, not simply different labels.
The third transition is from delivery to take-up. In the intensive group, 90.1% of firms completed the online platform, but 79.3% received individual mentoring and only 43.7% attended at least one masterclass in person. Recorded material was available online, so physical attendance was not the whole dose, but it was still a meaningful difference in exposure. In the online comparison group, 70% accessed the course, while only 37% took part in a quiz. Access was common enough to matter, yet completion and deeper engagement were much less complete.
These figures do not show that less-engaged founders lacked commitment. They show that “participation” is not a single condition. Founders cited time away from their businesses and travel distance as important reasons for missing in-person activity. A team may be working a second job, managing a production schedule, or serving early customers. A programme that demands attendance at a distant location may therefore screen for available time and geography as much as it delivers support. Virtual material can reduce that burden, but it may not provide the same interaction or accountability as individual work. The relevant measure is useful exposure, not the existence of a login.
Engagement is also shaped by perceived need. In the programme, firms with higher initial readiness were more likely to complete the first online phase but less likely to attend a masterclass. That pattern is consistent with a reasonable interpretation: a firm that already feels capable may see less value in a general workshop. The study cannot establish that this was the reason, but it makes the design issue visible. If a programme gives every startup the same lesson, the most prepared firms may disengage while less prepared firms may receive information without enough individual help to apply it.
This is where personalisation becomes a transition mechanism rather than a slogan. A programme has to diagnose what the startup is trying to change, then adjust the route, depth, and timing of support. The OECD warns that support mismatched to entrepreneurs’ needs can deter engagement or be counterproductive, and emphasises one-to-one coaching, targeted training, specialised programmes, and bespoke matchmaking. A startup cannot reach the next state if the route into the programme consumes more time than the problem it solves, or if the help arrives after the relevant decision has passed.
A useful dose is therefore the product of availability, relevance, and engagement. Remove any one of those and the nominal offer becomes weaker. Availability without relevance is noise. Relevance without availability is a promise. Availability and relevance without engagement is unused potential. None of these conditions guarantees an outcome, but each is necessary for the support to have a plausible chance of changing one.
The path from readiness to a business result
The next transition begins when the startup’s capabilities or evidence change. Readiness is often an intermediate state: the team can explain its proposition more clearly, measure its traction more credibly, understand its market better, or prepare more effectively for an investor conversation. These changes can matter. They can improve the quality of a decision or make a company easier for an outside actor to evaluate. But they are not identical to customers paying, financing arriving, the company surviving, or employment and revenue growing.
The difference is easiest to see as a chain. A startup may first become more capable of presenting its case. That may produce attention from an investor or customer. Attention may become a concrete step, such as a second meeting, a pilot, a financing negotiation, or work with an expert. The step may become a deal or a paying relationship. A deal may provide resources that support survival or growth. At every point, another actor makes a decision and another condition can intervene.
The World Bank study measured several of these states separately. Independent judges assessed investment readiness, while later follow-ups examined readiness actions, steps toward external financing, external financing received, operating status, and media attention. This separation is important. If a programme raises a judge’s score, the score tells us something about an immediate change in how the firm appears and performs in an evaluative setting. It does not tell us that the company has crossed the much longer distance to a durable business result.
The chain can also lose information at each transition. A coaching session may help the founders identify a customer segment, but the customer still decides whether the product solves a real problem. A stronger pitch may earn a meeting, but the investor still assesses the opportunity, terms, team, and market. Financing may arrive, but the company still has to spend it on a product or distribution system that works. The incubator can improve preparation and access while lacking control over the external decision that determines conversion.
This is the logic behind the “funnel of attribution.” The starting group consists of firms that apply or are admitted. A smaller group engages deeply. A smaller group becomes more ready for a particular opportunity. Some take steps toward financing or customers. Fewer complete a deal, survive, or grow. The narrowing is not only a statistical inconvenience. It describes a changing causal environment: as the startup moves outward from the programme, more of the result depends on customers, investors, competitors, regulation, timing, and the team’s own execution.
The loop matters. A stalled startup does not necessarily need to leave every programme; it may need a revised match, a different specialist, more time, or a decision to stop pursuing an opportunity. But the loop is not free. Repeatedly supplying the same support without updating the diagnosis consumes founder attention and programme resources. A credible incubator should be able to notice whether the startup is moving, not just whether it remains enrolled.
The same framework explains why an improvement in readiness can be predictive without producing a large treatment effect on final outcomes. A readiness score can contain useful information about a company’s team, technology, traction, market, and recent progress. Firms with stronger scores may later be more willing to take equity, more likely to adopt investor-oriented practices, or more likely to take steps toward financing. Yet the programme’s effect on that score may be too small, too uneven, or too distant from the binding constraint to create a reliably detectable change in financing or survival.
There can also be several valid outcomes that are not captured by a simple acceleration story. A programme may help a founder realise that the company is not ready to raise capital, preventing an expensive distraction. It may produce a rejection that exposes a product flaw early. It may help a team improve employment or technical capability without attracting venture capital. These outcomes should not be smuggled into a claim of growth, but neither should they be dismissed merely because a financing event did not occur. The outcome must be defined before the programme is judged.
That requirement is especially important when the programme’s headline result is attention. Media mentions, social-media activity, a place in a final pitch event, or an investor conversation can be useful signals. They may widen the set of future opportunities. But attention is an input to later decisions, not the decision itself. If the startup’s intended goal is customer revenue, then the relevant conversion measure must eventually include customer behaviour. If the goal is financing, it must include a deal or a clearly defined step toward one. A signal is not a substitute for the outcome it is meant to predict.
The chain also has implications for timing. A service can be useful at one stage and distracting at another. Pitch preparation may matter when the company has evidence and is ready to meet investors; it may be premature when the founders still need to test their market. International connections may matter once the offer is repeatable and the team can support a new geography; before that, they may spread attention. The OECD’s emphasis on evolving needs is therefore practical: the same startup can require different support as it moves through the chain.
The diagnostic conclusion is modest but powerful. When an incubator appears not to accelerate a startup, ask which transition failed. Was the company selected but not matched? Was the service promised but not delivered at a useful standard? Was it delivered but not usable within the founder’s time and geography? Did readiness improve without a relevant opportunity? Did attention fail to become a customer, financing, or operating result because another constraint remained? Each answer points to a different remedy, and none can be inferred from the word “joined.”
This chain gives us a sharper test for evidence. We need a comparison that tells us what the support changed, measures that do not replace final outcomes with convenient proxies, and enough follow-up to see whether the change survived contact with the outside world. The randomized experiment provides that kind of structure for one intensive investment-readiness offer. Its value lies not only in the result it reports, but in the distinctions it forces us to keep.