How MoorInnov maps the ecosystem

MoorInnov (moorinnov.com) positions itself as an atlas for practitioners, researchers, and policymakers, and explicitly invites readers to "cite our methodology". This note answers that invitation: it documents precisely what the platform measures, how the data is collected and transformed, and — crucially for responsible academic or policy use — what the platform does not yet measure.

It draws on a corpus of 27 reference articles and reports on the theory and measurement of entrepreneurial ecosystems, assembled for this project (see References), and applies their frameworks and cautions to the actual state of MoorInnov's data — without embellishing what has not yet been done.

MoorInnov adopts the now-standard definition in the literature (Stam, 2015): an entrepreneurial ecosystem is "a set of interdependent actors and factors coordinated in such a way that they enable productive entrepreneurship within a particular territory."

Stam and van de Ven (2021) operationalise this definition in ten systemic elements (formal institutions, culture, physical infrastructure, demand, networks, leadership, talent, finance, knowledge, intermediaries), linked by strong interdependencies, which produce entrepreneurial outputs (new firms, funding rounds) and, in turn, socio-economic outcomes (employment, value created).

An important methodological point highlighted by Wurth, Stam and Spigel (2022, 2023): most theoretical frameworks clearly distinguish the ecosystem's elements/conditions, its outputs, and its outcomes. In its current state, MoorInnov essentially maps the actors — an inventory of the entities that make up the ecosystem — and some of the elements (institutions, infrastructure, networks via actor type). It does not yet measure outputs (survival, growth) or outcomes (jobs created, value added). Nicotra, Romano, Del Giudice and Schillaci (2018) argue this distinction is essential to avoid conflating an inventory of actors ("eco-factors") with a measure of entrepreneurial performance ("eco-output").

The ten categories used by MoorInnov (startup, investor, accelerator, incubator, hub, university, corporate, public_agency, ngo, event_org) were chosen to match attributes identified in the literature as constitutive of an ecosystem (Stam & van de Ven, 2021; Spigel, 2017), rather than as an ad hoc nomenclature.

Each type maps to a systemic element. "startup" is the central entrepreneurial output the ecosystem is meant to produce (Nicotra et al., 2018). "investor" supplies capital and conditions the crossing of funding stages (Finance — Stam & van de Ven, 2021; Spigel, 2017). "accelerator" and "incubator" structure training, market access and the first network (Support / Leadership — Stam & van de Ven, 2021). "hub" is the physical space of co-presence and serendipity (Physical infrastructure). "university" produces human capital and mobilisable knowledge (Talent / Knowledge). "corporate" embodies market demand, intermediation, and sometimes investment or acquisition (Demand / Intermediaries). "public_agency" embodies formal institutions: regulation, labelling, public funding (Stam, 2015). "ngo" carries inclusion, advocacy, and supporting cultural structures (Culture — Welter et al., 2017). "event_org" densifies the network at points in time and diffuses cultural norms (Networks / Culture — Spigel, 2017).

As of this note, MoorInnov aggregates four corpora. The first two are public secondary sources collected by web extraction; the last two are official datasets obtained directly from Morocco's Digital Development Agency (ADD) via the data.gov.ma portal. No data has been collected through direct interviews with founders (see §9 on the gap with the text currently published on /about):

— OSE Connect — "The directory of support structures": a PDF directory of 604 structures (accelerators, incubators, universities, public agencies, coworking spaces…), extracted in July 2026. — start-up.ma — a community directory of Moroccan startups: all 65 listing pages parsed in July 2026 (~580 raw entries), filtered to entities actually based in Morocco, for 501 startups retained. — ADD Open Data — "BDD STARTUPS ADD 2024" (data.gov.ma): 1,028 raw entries from four combined official programs — "Jeune Entreprise Innovante" label (JEI 2024, n=500, see §7bis), Startup Hub Maroc (n=228), and the GITEX Africa Morocco 100 (n=100) and 200 (n=200) cohorts. After internal deduplication (merging entities appearing in several cohorts) and cross-referencing with the 501 startups already collected via start-up.ma, 646 new startups were identified. — ADD Open Data — "Données sur les porteurs d'initiatives" (July 2023): official list of 39 support structures (ESO) recognised by ADD. After cross-referencing with the 517 structures already catalogued, 24 new structures were identified and classified in the 10-type taxonomy (§3).

The first two sources are themselves third-party aggregators, not official registries; their accuracy has not been independently verified beyond the checks described in §5. The two ADD sources, on the other hand, come directly from a public agency and enjoy a higher confidence level — without, however, constituting an exhaustive census of the ecosystem (see §7bis on the actual scope of the JEI label).

— Text extraction (PDF → structured text; HTML pages → structured records). — Classification by type: source-site categories were mapped to the 10-type taxonomy; a manual review by entity name corrected 23 mis-classifications (mostly public agencies wrongly tagged as "support structures"). — Deduplication by normalised name. — City-level geocoding only (not exact address), via Google Places, with a small random jitter (~150 m) for map legibility — not to be interpreted as the real location of an office. — Documented exclusions: 18 entries with no identifiable city, 60 entries identified as non-Moroccan or without demonstrated activity in Morocco, 56 entries whose source type (media, research lab) does not match any of the 10 retained categories. — Cross-referencing with ADD datasets (July 2026): deduplication by normalised name against the 501 startups and 517 structures already catalogued; entries judged new receive an explicit provenance note in their description (e.g. "Listed by ADD: Morocco 200"), to trace the official source without conflating it with an independent verification by MoorInnov itself.

Geographic concentration. The over-representation of Casablanca and Rabat reflects both a real, well-documented concentration in African ecosystems (see Herrington & Coduras, 2019 on sub-Saharan Africa) and a coverage bias of the two source directories, themselves more active in those two cities.

Currency of status. "Active" status reflects the source's last update, not an independent verification; cessation dates are often missing.

Absence of funding data. MoorInnov does not yet distinguish well-funded actors from others, nor local from foreign capital — a significant gap given Colonnelli, Cruz, Pereira-Lopez, Porzio and Zhao (2026), who find that around 80% of venture deals in Africa involve a foreign investor and more than 60% of funded founders have studied or worked outside the continent — a pattern MoorInnov cannot yet test for the Moroccan case.

Absence of relational data (as of today). Who invested in whom, who accelerated whom: this layer does not yet exist in the collected data. The method for computing "connectivity" once these data are available is documented in §11.4; it remains unmeasurable until the relationships table is populated.

Absence of probabilistic sampling. Unlike the Global Entrepreneurship Monitor, which surveys representative adult samples (Reynolds et al., 2005; Bosma, 2013), MoorInnov is a directory of actors who have made themselves visible (or been referenced) on specialised websites — a self-selection bias that likely favours digitally visible, urban, and francophone/anglophone organisations, at the expense of informal or rural entrepreneurship.

Language coverage. Sources are in French and English; structures referenced only in Arabic or Amazigh are likely under-represented.

For responsible academic or policy use, it helps to situate MoorInnov relative to existing standards for measuring entrepreneurial ecosystems:

— It is not a complete official "Startup Act" registry. Unlike Tunisia (Sold, 2018; Ali, Calì and Rijkers, 2025), Italy (Menon et al., 2018) or Nigeria (Stever, 2023), Morocco has no broad legal status granting extensive tax exemptions to labelled startups. There is however a narrower scheme, detailed in §7bis, which MoorInnov can now partially cross-reference. — It is not a representative GEM-style survey (Bosma, 2013; Reynolds et al., 2005). — It is not (yet) a validated ecosystem-quality index. Leendertse, Schrijvers and Stam (2022) and Stam et al. (2026) build composite indices from multiple cross-checked data streams (business registrations, surveys, patents, regional GDP). The future MoorInnov Ecosystem Vitality Index (§11) will only be robust once funding and relationship data are actually collected. — It is, however, a systematic, reproducible, continuously-correctable inventory of actors — closest in spirit to the Startup Cartography Project (Andrews et al., 2022), which similarly combines administrative records to produce a public map, though at an earlier stage of data-verification maturity.

Version 1.0 of this note stated that Morocco has no statutory registry of labelled startups. That statement needs qualification. Since 18 September 2019, ADD has issued an official "Jeune Entreprise Innovante (JEI) en nouvelles technologies" label, in partnership with the Foreign Exchange Office, OMPIC, the Central Guarantee Fund, CGEM, APEBI and GPBM. The label mainly grants an easing of foreign-exchange controls (up to MAD 1,000,000 per year by international card for the purchase of digital services abroad), rather than a broad tax regime comparable to the Tunisian Startup Act. 398 firms had been JEI-labelled between 2019 and September 2023.

The ADD dataset integrated into MoorInnov in July 2026 (§4) includes 500 entities tagged "JEI 2024". MoorInnov cannot however guarantee that this tag reflects an active JEI status at the moment of consultation (the label has a limited term and may be withdrawn): it is treated as a declarative provenance mention, on par with other source tags (Morocco 100/200, Startup Hub Maroc), and not as a verification of currently-valid legal status.

According to Baumol (1990) and Lucas and Fuller (2017), entrepreneurial activity is not intrinsically beneficial: its social value depends on the institutional context and on the allocation of entrepreneurial effort between productive, unproductive, or even destructive activities. Scott (2025) goes further, arguing that an ecosystem can, in the absence of safeguards and shared virtue, incentivise the "dark side" of entrepreneurship.

The presence of an entity in the MoorInnov atlas is therefore descriptive, not evaluative: it signals that an entity exists and identifies itself (or is identified by a third-party source) within the ecosystem taxonomy — not that its activity creates net social value, nor that it meets any legal, ethical or governance standard. This caveat should appear explicitly on the site, in particular for policymakers who might be tempted to read presence in the directory as a quality label.

The moorinnov.com/about page currently states that "each actor is reviewed by ecosystem contributors before publication", that "records combine primary interviews with founders, public regulatory filings, and open data", and that "capital and headcount data are refreshed quarterly". None of these three statements match, at this stage, the pipeline actually implemented (§4–§5): there is no contributor-review circuit yet, no primary interviews, no systematic regulatory filings consulted, and no automated refresh cadence.

Recommendation: replace this text with a version aligned to this note (for example by pointing directly to this downloadable document, as the reference platforms do — Andrews et al., 2022; Stam et al., 2026), announcing the three missing elements as a roadmap rather than an accomplished fact. This is an important credibility point vis-à-vis the researcher audience explicitly targeted by the site.

Current state: one-off extraction (July 2026), with no automated refresh. The "Submit an actor" form exists for community correction, but submissions do not yet follow a formalised verification circuit. This state must be communicated explicitly until a real review flow is in place.

This section documents, for the first time in full, the computation method for the future regional Ecosystem Vitality Index. It replaces and details what was previously only an intention.

11.1 Framework and justification. The index follows the Density–Fluidity–Connectivity–Diversity (DFCD) framework of the Kauffman Foundation (Bell-Masterson & Stangler, 2015) — the reference framework for measuring the vitality of an ecosystem (as opposed to static indices that only count institutions), extended with a fifth dimension, Transparency (see 11.6), justified by Johnson, Hemmatian, Lanahan and Joshi (2022). Three methodological choices structure the whole calculation: (1) geometric-mean aggregation, never arithmetic — because Stam and van de Ven (2021) show that ecosystem elements are strongly interdependent: a weak link must penalise the whole, not be compensated by a strong one elsewhere ("Penalty for Bottleneck" method — Almeida & Daniel, 2025); (2) min-max normalisation recomputed at each period, across all active regions, so that the scale remains comparative as data grow; (3) a 12-month rolling window recomputed monthly, to smooth out shocks — consistent with Belfanti, Riva and Alberti (2025).

11.2 Density. Stock measure: the number of active actors, net of closures, per 100,000 inhabitants when regional population is known; otherwise a raw count with reduced data_coverage.

11.3 Fluidity. Flow measure — the vitality of an ecosystem shows in its renewal, not only its size. Four normalised, then averaged signals: creation rate (new entities over total active), funding velocity (funding_rounds over active startups), stage progression (seed → series A, etc.), and diaspora inflow (new entities with diaspora = true) — a talent return signal in the sense of Stam and van de Ven (2021) and of Audretsch, Fiedler, Fath and Verreynne (2024) on geographically unbounded ecosystems.

11.4 Connectivity. The relational signal, justified by Johnson, Hemmatian, Lanahan and Joshi (2022). Two components: network density (active relationships involving at least one actor from the region, divided by the maximum possible pairs) and average investor syndication per funding_round (co-investment is itself a connectivity signal).

11.5 Diversity. Shannon entropy on categorical fields of the region's active actors — sectors, type, and stage — the normalised mean of the three entropies (0 = homogeneous, 100 = perfectly diversified). High sectoral and stage diversity reduces the ecosystem's dependence on a single sector or maturity cohort.

11.6 Transparency (new dimension). Added to measure governance maturity and trust in the ecosystem — an indirect signal of institutional quality, in the spirit of Acemoglu and Robinson (2012). Theoretical justification: an ecosystem is only measurable to the extent that its actors voluntarily disclose their data (Johnson et al., 2022), and what distinguishes an entrepreneurial ecosystem from a classical industrial cluster is precisely the presence of voluntary knowledge spillovers between actors (Autio, Nambisan, Thomas & Wright, 2018). Four sub-components, aggregated with a geometric mean: profile completeness (mean % of non-null fields), financial disclosure rate (funding_rounds with disclosed amount), verification rate (funding_rounds + relationships marked "verified"), and engagement score (average of claim rate, self-maintenance rate, and open-data participation). Important: as recalled in §8, Transparency does not measure the intrinsic quality of an actor, only the quantity and reliability of data available about it. This score depends directly on the "Claim this listing" mechanism — without it, it will remain close to zero indefinitely.

11.7 Composite aggregation and handling of missing data. Composite score = geometric mean of the five scores (Density, Fluidity, Connectivity, Diversity, Transparency), with equal weight (1/5) by default — no differentiated weighting is applied without explicit justification in the interface (consistent with the §8 caveat on the non-evaluative character of the atlas). Each score also records a data_coverage field (% of non-null source fields). A region where funding_rounds is empty must display a Fluidity sub-score flagged "low confidence", never a bare 0 indistinguishable from a genuinely low score. Two regions must not be compared if their data_coverage differs by more than 30 points without an explicit visual warning.

11.8 Roadmap. Once the funding_rounds, relationships and actor_claims tables are effectively populated, this method must be cross-checked against Stam, Nkontwana, McDonald, Murenzi, Addo, Bayuo, Baah, Riezebos and Gelissen (2026), who propose an Africa Entrepreneurial Ecosystem Index with 21 indicators across 7 dimensions — the closest reference to the Moroccan context. The methodological sequence documented by Crotti, Potter, Shah and Stam (2025) for the OECD (conceptual framework → indicator choice → normalisation → aggregation → imputation of missing values → robustness tests) remains the reference to follow for validating and refining the formulas above once real data are available across several periods.

This section documents a planned MoorInnov extension: a per-actor relational view showing who collaborates with whom, who is linked to whom (investment, mentorship, alumni), and, where applicable, publicly disclosed family ties.

12.1 Theoretical anchor. Spigel (2017) has long formalised that an entrepreneurial ecosystem is fundamentally relational, not a mere list of side-by-side actors. Social networks — mentorship, co-investment, talent mobility across organisations — matter as much as the material or cultural resources listed in earlier sections. Three additional refinements justify including publicly disclosed family ties alongside professional ties: Benavides-Salazar, Iturralde & Maseda (2021) show that entrepreneurial families shape ecosystem development through family social-capital mechanisms that cross firm boundaries; Ge, Stanley, Eddleston & Kellermanns (2019) show that in emerging markets — a context directly relevant to Morocco — family ties frequently compensate for weak formal institutions; Arrégle et al. (2015) find the relationship between family-tie density and venture growth is curvilinear, not linear — beyond a threshold, high family-tie density can hurt as much as help. The map must therefore stay strictly descriptive, never presented as a quality score, consistent with §8.

12.2 Data schema. Three new tables: people (individuals — founders, executives, mentors), affiliations (person↔organisation link, with role and dates), and people_relationships (person↔person link — co-founding, mentorship, alumni, and family ties). The relationships table already planned for the Vitality Index (organisation↔organisation) remains unchanged and coexists with this new schema.

12.3 Collection method. Scraping LinkedIn or any social network is explicitly excluded (terms-of-use violation, legal risk). Three legitimate channels are used, in decreasing order of reliability: self-declaration, via an extension of the “Claim this listing” mechanism (§11.6); targeted press research on priority actors, with a strict rule that a family tie is only recorded when a source reports an explicit public statement by the person concerned — never inferred from a shared surname; and official funding announcements, which almost systematically name founders and participating investors — the most reliable source for populating founder / general_partner affiliations.

12.4 Ethical and privacy guardrails. Mapping family ties means handling sensitive personal data about identifiable people. Four non-negotiable rules: no inference of a family tie from a shared surname, a photo, or any other indirect deduction; an unverified family tie is not shown publicly by default — only once sourced and the person's public statement identified; any person concerned (or their heirs) may request removal of a family tie about them, without justification — in line with Moroccan Law No. 09-08 on the protection of individuals with regard to the processing of personal data, supervised by the CNDP; a documented family tie signals neither favouritism, conflict of interest, nor value judgement — the same descriptive reservation as the rest of the atlas (§8) applies and is displayed explicitly next to any shown tie.

12.5 Building the map. For a given actor, the map combines direct organisation↔organisation links and links mediated by people affiliated with it (their other affiliations, their co-founding, mentorship or family relationships, and the affiliations of those linked people). Depth is capped by default at two degrees of separation: beyond that, both graph readability and data reliability degrade quickly.