Ework's Five-Year Margin High Is a Farewell, Not a Milestone
Q3 revenue fell 8% but gross margin hit 4.3% — then management spent the call explaining why it will now trade that margin for volume, with AI on both sides of the ledger.
EWRK.ST · Earnings Call · 2026-10-22
The margin peak nobody wants to own
Ework's third quarter reads like a company that has spent two years defending its income statement and has now decided to spend it. Revenue fell roughly 8% to SEK 2.8bn from SEK 3.0bn last year, with consultant volumes down in every market bar Norway. Yet gross profit slipped only 7%, because gross margin rose to 4.3% — the highest level in five years. EBIT landed roughly in line with last year, EPS grew 14%, and net financial items swung from a SEK 6.9m cost to a SEK 4m credit on lower interest and favourable currency.
That combination — falling revenue, rising margin, flat profit — is usually sold as resilience. Ework's CFO instead spent her prepared remarks pre-emptively dismantling it: “We have reported before and mentioned today that volume is our top priority and that that might impact the gross margins going forward. However, long client contracts create resilience in gross margins, and this effect will take time.” — Johanna Estra, CFO · 2026-10-22 Pressed again in Q&A, she repeated it: “we are focusing more on building volumes again, and that might lead to lower gross margins going forward. But the cost discipline will remain.” — Johanna Estra, CFO · 2026-10-22
This is the crux of the report. The 4.3% margin is not a platform; it is a peak that management has already told you it intends to give back. The mechanism is explicit: business volumes are the new top priority, and the company just finished designing a strategy that assumes margin pressure. A shareholder who reads only the margin line reads the story backwards.
AI: the fastest-growing competence and the thing that eats the model
The genuinely new material is AI. Management reports that AI-related demand rose 134% over the rolling twelve-month period, making it the company's fastest-growing competence area. More than one in six new consultant profiles joining the network now carry AI skills, up 11% versus Q2. The demand mix is maturing quickly — from chatbot development and LLM training toward agent development, retrieval-augmented generation and vector search. There is real substance here: Ework sits between clients and a fragmented supply of niche specialists, and the CEO is candid that this is structurally where the expertise lives. “Connecting clients with specialists is what we have always done. Cutting-edge knowledge within the field of AI increasingly sits with specialists who are freelancers who work in niche consultancies.” — Operator, Operator · 2026-10-22
The tension is that the same technology is simultaneously the company's deflation problem. The 2027–2029 strategy is built around expected effects of AI: automation of administrative work, a lower cost to serve, multiple service tiers, and a push into statement-of-work and outcome-based delivery. The automation potential is what allows Ework to pursue smaller clients profitably for the first time — but it is also what management cites when warning about margin. AI is not an add-on story here; it is the investment case and the risk factor wearing the same coat.
Against the market's dominant themes, this is a distinct angle. The global keyword board is wall-to-wall AI data center, compute, memory and power — the hardware and infrastructure trade. Ework has essentially zero exposure to that. Its AI trade is the labour-substitution trade: demand for AI competence rising alongside pressure on the consulting model that delivers it. Notably, the company's own keyword history until now had no AI thread at all; "AI competence", "new AI initiative" and "experimentation to implementation" all appear for the first time this quarter. This is a genuinely new pillar, not recycled narrative.
The pipeline exists; the conversion date does not
Order intake fell about 9%, with the largest declines in Denmark and Poland and the automotive and consulting verticals still weak. But the soft indicators point the other way: requests in Q3 were up 12%, activity picked up after the summer and into September, Sweden's decline slowed, Norway grew 21% on telecom and public sector, and the sales pipeline expanded. The company also won a notable mandate — an exclusive MSP agreement for interim consultants across all competency areas in Sweden with Tetra Pak — plus renewals with Husqvarna, SKF, Ellevio, Malmö Stad, Öresundskraft and Telemark Fylkeskommune, and new business at Nobia.
The problem is the shape of the conversion curve, and management was unusually guarded about timing. Asked directly when improved activity would fall into orders: "No comments." Asked again: "No comments." The eventual answer was “if this activity continues, which is really positive, of course it will have a gradual effect, but it's hard to say, and also back to the lead times, how fast that will go.” — Johanna Estra, CFO · 2026-10-22
lead times is a newly prominent keyword for this company, and the CEO's explanation is structural rather than cyclical: cultivating a client takes years, bids are formal, awards take months to activate. The CEO also conceded that Ework's broker model is more macro-levered than a pure consultancy — “the broker role that Ework has is more subjected to macro trends than pure consultancies. At least that's my take on it. On the flip side, that should also work in our favor on the way up.” — Daniel Almgren, CEO · 2026-10-22 That is both the bull case and the honest hedge.
There is a sector signal worth flagging. Peer staffing reporters this week struck a more constructive note on labour demand — Randstad's team pointed to incremental demand returning to Labor markets, and Robert Half reported PERM placement revenue up 6% against a low unemployment backdrop. Ework is adamant it is not yet in that camp — “it is still too early to talk about a broader market recovery and lead times remain long.” — Daniel Almgren, CEO · 2026-10-22 If the staffing recovery is real, Ework is the late-cycle derivative of it, levered to a rebound it cannot date.
A new scoreboard, and a quiet admission
Buried in the Q&A is arguably the most telling line of the call. Asked what investors should measure, the CEO effectively retired EBIT: “our primary focus is on absolute GP going forward... the ratio between GP and net result.” — Daniel Almgren, CEO · 2026-10-22 When a management team changes its yardstick in the same quarter it flags margin give-back, it is telling you the old yardstick will look worse.
The offsetting levers are real but slow. The reorganisation delivers SEK 18m of annual savings, plus SEK 5m from facility reductions, with full effect only in 2027. Germany was addressed via an acquisition granting an AÜG licence; network expansion targets Poland, Germany and Benelux. Denmark remains the sore spot — down 25% — and the CEO's diagnosis is client concentration, not market: “in both Denmark and Norway... they have a high degree of client concentration... We will only get out of that dependency on a few select accounts when we establish a broader client base.” — Daniel Almgren, CEO · 2026-10-22 That is a years-long fix dressed up as a regional problem.
The verdict
Ework delivered a quarter that looks better than it is and sounds better than it will be. The margin is a peak; the volume recovery is a pipeline, not a P&L line; the AI franchise is genuinely the fastest-growing part of the business and genuinely a threat to the pricing model underneath it. With a market capitalisation near SEK 900m, this is a small-cap execution story now fully dependent on converting a 12%-higher request book into signed, activated volume before the margin it just gave back starts to bite. Management says the thinking phase is over. The next two quarters are the proof.